Game Changers

power solutions 2010 Issue 03 2010 Issue

Applications Your guide to maximizing IT efficiency dell.com/powersolutions 03 “Unlimited” Oracle – Siebel – PeopleSoft – JD Edwards Game changers: How blade servers up the ante on data center efficiencyGame changers: How blade servers Game changers How blade servers up the ante Continued New Releases on data center efficiency

Customer Driven Product Roadmaps

Dedicated Development Teams

No Forced Migrations

Get better results with proven applications tailored for your industry, processes and geography.

Dell blade customers Building the dynamic exhibit cutting-edge agility data center

oracle.com Smooth blade management Desktop power management or call 1.800.ORACLE.1 using CMC 3.0 via the Dell KACE appliance

DPS-201003-STD Cost-effective tiering for Seamless Microsoft Exchange virtualized iSCSI SANs across data centers

Copyright © 2006, Oracle. All rights reserved. Oracle, JD Edwards, PeopleSoft and Siebel are registered trademarks of Oracle Corporation and/or its affiliates. Other names may be trademarks of their respective owners. Can$19.95 US$18.95

PUB NOTE: Please use center marks to align page. CUSTOM Job No.: 001592 Live: 7.875” x 10.375” APPROVALS BY DATE PRODUCTION NOTES READER File Name: EBS_AppsUnlimited_1592_DPS Trim: 8.375” x 10.875” Traffic Product: Applications Bleed: 8.625” x 11.125” Production 01 8.375” x 10.875” Headline: Applications “Unlimited” Color: 4C LASER% Proofing Date: Fri, Jul. 23, 2010 1:45 PM Production: Graphic Mgr. Pub: Dell Power Solutions Dell Power Solutions Fonts: Traffic: HQ Adv. Mgr. Univers LT Std. Font Family RELEASED Magazine Library Ref.: 000 Buddy Check Please examine these publication materials carefully. Any questions regarding the materials, please contact Darci Terlizzi (650) 506-9775 SQLSrvr_Ocean.indd 1 5/20/10 5:04:18 PM 2010 Issue 03

8 cover story Game changers By Robert Bradfield, Chris Christian, and Jeanne Feldkamp

Designed for unparalleled computing density, Dell™ blade servers boost flexibility and minimize costs with innovative power and cooling, intelligent network integration, seamless manageability, and exceptional reliability. How nimble is your IT?

Simple Mail Transfer Protocol Remote users (SMTP) servers

IMAP4, POP3, HTTP, RPC over HTTP, MAPI over VPN Internet Switched Gigabit Ethernet or 10 Gigabit Ethernet network

Data center 1 Internal users Internal users Data center 2

F5 BIG-IP systems F5 BIG-IP systems Member 1 Member 2 Member 3 Member 4 Demilitarized DMZ zone (DMZ) Global Trac Manager Global Trac Manager Production applications Backup and archive Local Trac Manager Local Trac Manager volumes volumes Edge Transport Edge Transport servers servers Microsoft Active Microsoft Active Application storage pool Default storage pool Directory® server Directory server LAN LAN WAN

F5 BIG-IP systems F5 BIG-IP systems

Smart tiering Fast access to End-to-end OptimizLocal Trac Managered WAN optimizationEfficient, high-Local Trac Manager 24 54 64 72 Application Security Manager 79 Application Security Manager for virtualized high-demand services to WebAcceleratorExchange™ module trafficWAN Optimization Module density expansionWebAccelerator WAN Optimization Module iSCSI SANs application data simplify Oracle® spanning chassis for GPU Client Access servers Mailbox servers (DAG) Client Access servers Mailbox servers (DAG) environments multiple sites computing Dell PowerEdge™ servers

™ Dell EqualLogic storage Dell PowerEdge blade servers Dell EqualLogic storage

Dell Agile Consolidated model (1,000 mailboxes) Dell Agile Consolidated model (virtualized, 10,000 mailboxes)

Blade solutions for the Efficient Editor’s comments 6 @Socially speaking Data Center By Tom Kolnowski

Storage optimization 24 Cost-effective tiering for virtualized iSCSI SANs By Achmad Chadran and Keith Swindell

28 Streamlining enterprise database environments for business continuity By Annette Cormier, Jason Kotsaftis, and Jeff Browning 16 Cutting-edge 20 Smooth operator agility By Roger Foreman 36 Accelerate operations with next-generation storage efficiencies What do a wind farm developer, a Built by implementing an innovative, By Annette Cormier, Eric Cannell, clothing manufacturer, a maritime user-driven development process, the and Brad Bunce security and tracking services provider, Dell Chassis Management Controller 40 Best practices for deploying iSCSI and three institutions of higher learning (CMC) 3.0 embedded management tool SANs in DSS database solutions have in common? They heightened IT enables administrators to simply and By Suresh Jasrasaria, Ananda Sankaran, efficiency, reliability, and performance intuitively manage multiple blade servers and Chris Almond using Dell blade servers. from a single console.

32 Storage optimization Taking an intelligent data management approach to archiving data

By Kay Benaroch and Joe Colucci

By employing a strong archiving strategy as the foundation for an intelligent data management program, enterprises can enhance efficiency for IT staff, meet compliance requirements, and streamline backup processes.

By M. Consuelo Ortiz

As a key component of the Dell Efficient Data Center strategy, the Dell Virtual Integrated System architecture helps organizations move from a static to a dynamic data center and increase efficiency in the virtual era by effectively managing technology and the people and processes that manage the technology.

45 Rightsizing storage for small to 72 Designing seamless Microsoft midsize virtualized environments Exchange deployments across Systems management By John Mannix and Kyle Walczak data centers 90 By Kong Yang, Jeff Sullivan, and 48 Robust data protection in a Fred Johnson virtual world Deploying By Ed Casmer 76 A smart approach to application intelligence automated power Virtualization By Patrick Sweeney and management for 54 Using multitiered arrays to optimize Matthew Dieckman application performance PC environments By Andrew Gilman and Suresh Jasrasaria High-performance computing 79 Expanding the boundaries of Enhanced plug-and-play 58 How desktop virtualization enhances GPU computing power management can end-user productivity accrue significant savings By Todd Mitchell and Doug Coombs 84 Optimizing HPC clusters with 10 Gigabit Ethernet iWARP and help IT staff efficiently 60 Successful liftoff: Achieving cloud technology control the PC infrastructure. computing in the enterprise By Tom Stachura The Dell KACE™ K1000 By Lilac Schoenbeck management appliance Systems management offers this capability in Services 88 Managing the layers of a virtualized a flexible, easy-to-use 64 How end-to-end services can environment transform Oracle environments By Ed Casmer package. By Tim Daigle and Thomas Kopec 93 Simplifying custom Microsoft Networking Windows deployments 68 Enhancing Ethernet network By Gong Wang and Wayne Liles controllers for server virtualization By Dhiraj Sehgal, Abhijit Aswath, and Srinivas Thodati

Browse our Web site: Get the Digital Edition: Follow us on Twitter: dell.com/powersolutions powersolutionsdigital.dell.com twitter.com/powersolutions online exclusive Power Solutions Digital Edition Maximizing replication efficiency Dell Power Solutions Digital Edition offers a comprehensive IT library at your in Oracle environments fingertips—featuring powerful searches across archived issues and direct links to By Annette Cormier and Jason Kotsaftis related resources. For instant accessibility, visit powersolutionsdigital.dell.com. In Oracle database environments, Dell/EMC storage technologies help simplify and enhance

the time-consuming replication process while Feature section Blade solutions for the Efficient Data Center enabling application-level consistency and RES: Running like the wind on understand its needs, and recommended Dell blade servers an energy-efficient HPC cluster built on For rapidly growing wind farm developer 16 Dell PowerEdge M610 blade servers with recovery to previous points in time. Renewable Energy Systems (RES), a power-efficient Intel® Xeon® X5550 processors running HPC platform is mission critical. Having built Platform Cluster Manager – Dell Edition software. more than 80 wind farms around the globe, RES Mounted in a PowerEdge M1000e modular continues to expand into new regions, including blade enclosure and connected with Dell Turkey and South Africa, and had to upgrade its PowerConnect™ M6220 switches, the cluster existing Dell HPC solution to quickly and accurately easily fits in the space available, and includes Powerful assess these complex emerging markets. a Dell PowerVault™ MD1000 direct attach array performance

Minimizing its carbon footprint and saving for high-volume storage. In this video, learn more Cutting-edge space were key requirements, particularly because “Now we can complete the job in two to about how Dell blade servers an RES sister company was hosting the system three weeks—approximately 20 times faster— helped RES meet its need for a high-performance, until a new data center could be built. Because which means we can get wind farms in place energy-efficient cluster. agility IT resources were scarce, simplified systems quicker,” says Peter Stuart, technical manager at management was also essential. RES. In addition, the company has reduced power youtube.com/watch?v= cwkDh-_pVnk The Dell Global Infrastructure Consulting consumption by approximately 75 percent and Dell Power Solutions Magazine and special Services team worked closely with RES to freed up valuable IT staff time.

What do a wind farm developer, a clothing manufacturer, edition articles are also available online at Cornell University: Simplifying platform and DataDirect Networks storage. a maritime security and tracking services provider, and HPC solutions on the TeraGrid Force10 Networks switches were sourced HPC is the backbone of the work performed through Dell. This HPC cluster allows a wide three institutions of higher learning have in common? by the Cornell University Center for Advanced range of scientific researchers and students to dell.com/powersolutions. Check the They heightened IT efficiency, reliability, and performance Computing (CAC), which supports Cornell use MATLAB to help reduce the time to solution researchers and students from scientific in a seamless experience, without having to ™ using Dell blade servers. disciplines across the United States. The tackle the complexity and intricacies of parallel center recently received a National Science computing. For example, external researchers Dell Power Solutions Web site for early-release Foundation grant to deploy the MATLAB were able to remotely access all 512 cores of the fficiency has become the watchword across virtually every numerical computing environment to a cluster MATLAB cluster to model a hepatitis C virus, a called “MATLAB on the TeraGrid.” The goal was major cause of liver disease worldwide. line of business. The good news is that technology advances to demonstrate a working model for high- “We’re confident that scientific users and in today’s blade server infrastructures make it a snap to performance utility computing with MATLAB, engineers from across the nation will benefit from articles, how-to’s, case studies, best practices, and to encourage other software vendors to the reliability and performance of the Dell blade optimize data center efficiency and manageability—enabling E develop similar capabilities. solution while, on the IT side, Cornell will benefit organizations around the world to invest newfound IT cost savings The cluster uses Dell PowerEdge M600 blade from reduced power and space requirements and in initiatives that inspire innovation and accelerate growth. servers with Intel Xeon 5420 processors running the ability to maintain the blades with in-house and expert tips you won’t find anywhere else. the Microsoft® Windows® HPC Server 2008 staff,” says CAC director David Lifka. The success stories highlighted here represent a small but diverse sample of how organizations are putting Dell PowerEdge™ blade servers to work in ways that help maximize IT efficiency “We’re confident that scientific users and and flexibility, whether to consolidate data center resources, engineers from across the nation will benefit boost savings on power and cooling, virtualize servers for rapid from the reliability and performance of the deployment, or process data-intensive applications on a high- Dell blade solution.” performance computing (HPC) platform. Their results may — David Lifka surprise you. Director at Cornell CAC Dell blade servers optimize data center April 2010 efficiency—and boost organizational agility— through exceptional computing density, energy efficiency, network integration, and manageability

16 2010 Issue 03 | dell.com/powersolutions dell.com/powersolutions | 2010 Issue 03 17

Efficient Enterprise framework Take control of your data center with open, capable, and affordable solutions that help you focus on driving business innovation rather than just “keeping the lights on.” For the full story, visit dell.com/efficiency.

Index to featured Dell customers Index to advertisers Join the Dell TechCenter community at Antioch Unified School District.....91 Advanced Micro Devices, Inc...... C3 Cornell University Center for American Power Conversion delltechcenter.com to share your experiences, Advanced Computing...... 17 Corporation...... 67 voice your concerns, and pick the brain of your favorite National Center for BMC Software, Inc...... 23 Supercomputing Applications... 82 Broadcom Corporation...... 5 Dell expert—or be the expert by signing up for the Pacific Sunwear...... 18 Extreme Networks, Inc...... 15 Dell Masters program. And you won’t want to miss Pole Star Space Applications...... 19 F5 Networks, Inc...... 7 TechTuesday chats, community forums, discussion groups, Renewable Energy Systems...... 17 Intel Corporation...... 87 Stony Brook University...... 19 Microsoft Corporation...... C2 blogs, white papers, video demos, and much more. Thomas College...... 18 Network Engines, Inc...... 35 Xtium...... 56 Oracle Corporation...... C4 SonicWALL, Inc...... 83

4 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Broadcom® iSCSI HBA on Dell™ PowerEdge™ Servers with VMware® vSphere™

Broadcom best-in-class 10 GbE and 1 GbE iSCSI HBA and boot solutions provided “inbox” in VMware vSphere 4.1 for Dell PowerEdge servers

Contact your Dell sales representative for more information

©2010 Broadcom Corporation. Broadcom®, the pulse logo, Connecting everything®, and the Connecting everything logo are among the trademarks of Broadcom Corporation and/or its affiliates in the United States, certain other countries and/or the EU. Any other trademarks or trade names mentioned are the property of their respective owners. ITRK29991 Editor’s comments

Editorial staff Editor-in-chief and publisher Tom Kolnowski Managing editor Debra McDonald Features editor Kathryn White Associate managing editor Jim Duncan Senior editors Dawn Davidson, James Hurd, and Terrence O’Donnell Editorial assistant Amy J. Parker Art director and cover designer David Chan Designer and illustrator Cynthia Webb Business development manager Cissy Walker @Socially Staff writers Romy Bauer, Jeanne Feldkamp, Julie Jervis, Greg Thomas, and Chris Young

Contributing writers Chris Almond, Abhijit Aswath, Kay Benaroch, Robert Bradfield, Jeff Browning, Brad Bunce, speaking Eric Cannell, Ed Casmer, Achmad Chadran, Chris Christian, Joe Colucci, Doug Coombs, Annette Cormier, Tim Daigle, Matthew Dieckman, Roger Foreman, Andrew Gilman, Suresh Jasrasaria, Fred Johnson, Thomas Kopec, Jason Kotsaftis, Wayne Liles, John Mannix, Todd Mitchell, M. Consuelo Ortiz, Ananda Sankaran, Lilac Schoenbeck, Dhiraj Sehgal, Tom Stachura, Jeff Sullivan, Patrick Sweeney, Keith Swindell, Srinivas Thodati, Kyle Walczak, Gong Wang, and Kong Yang lame it on the allure of social media chatter. TweetDeck, my favorite desktop Contributing photographers Tony Bolding, Bryan Kuntz, application for monitoring Twitter, was dutifully displaying tweet after tweet Adran Matte, Joey Pena, and Bryce Vickmark originating from the eminently mobile attendees at the recent VMworld 2010 Advertising sales Sales director Kyle Walkenhorst (323-340-8585) conference in San Francisco. But with our press deadline for this issue looming National sales manager Shaun Mehr (949-923-1660) Western U.S. and South/Central America sales Melany Galley Blarge, the editing process for pages in my workflow queue had unfortunately slowed to a crawl. (949-481-1125) That was a small price to pay, given the fact I never had to leave the Dell offices Eastern U.S. sales Steve Branda (201-483-7871) Canada, EMEA, and APJ sales Mark Makinney (805-709-4745) in Austin to socially experience the flavor of VMworld, and the value provided by the Advertising assistant Scott Hallquist (323-254-0905) Ad coordinator Kathy Hartlove real-time nature and richness of many tweets from the individuals I routinely follow on Reader services Twitter. Among the more memorable 140-character bursts were instant knowledge and Subscriptions are complimentary to qualified readers

® who complete the online subscription form. For the Print links about the VMware vCloud Director announcement from the keynote address, Edition: To sign up as a new subscriber or change your and a stream of information and photos around the Dell™ Virtual Integrated System (VIS) existing subscription, access the Subscription Center at dell.com/powersolutions. For the Digital Edition: architecture from the Dell Super Session. To sign up as a new subscriber or convert your Print Edition subscription to the Digital Edition, visit Socially closer to home, the Dell Power Solutions editorial team is also on Twitter, powersolutionsdigital.dell.com. For other subscription services, please e-mail [email protected]. where we are tweeting many new articles and exclusive online-only postings as soon as About Dell they emerge from our content factory, along with news items we feel may be of interest. Dell Inc., headquartered in Round Rock, Texas, near Austin, You can follow us at twitter.com/powersolutions (@powersolutions). In addition, many listens to its customers and delivers innovative technology and services they trust and value. Uniquely enabled by its of our contributing authors and colleagues, as well as Dell product and solution teams direct business model, Dell is a leading global systems and services company and No. 34 on the Fortune 500 list. For we interface with, are frequent participants on Twitter. Here is a starter list of individuals or more information, visit our Web site at dell.com.

teams you may want to consider following: Susan Beebe (@susanbeebe), Dell HPC team Dell cannot be responsible for errors in typography or photography. Dell, the Dell logo, Dell KACE, Dell OpenManage, (@HPCatDell), Dell Inside IT blog (@DellInsideIT), Dell Storage team (@dell_storage), Dell Dell Precision, EqualLogic, PowerConnect, PowerEdge, and KACE team (@DellKACE), Franklin Flint (@franklinatdell), Andrew Gilman (@andrewgilman), PowerVault are trademarks of Dell Inc. Other trademarks and trade names may be used in this publication to refer to either Scott Hanson (@dellservergeek), Dylan Locsin (@DylanAtDell), Marc Malotke (@MarcDELL), the entities claiming the marks and names or their products. Dell disclaims any proprietary interest in the marks and names Lionel Menchaca (@LionelatDell), Jeff Sullivan (@SANpenguin), Greg White (@gregorydwhite), of others.

and Kong Yang (@kongy_dell). Dell Power Solutions is published quarterly by Dell Inc., Dell Power Solutions, One Dell Way, Mail Stop RR5-05, Round Our primary focus at Dell Power Solutions Magazine, of course, continues to be Rock, TX 78682, U.S.A. No part of this publication may be reprinted or otherwise reproduced without permission from the delivery of in-depth content designed to help you increase the efficiency of your IT the editor-in-chief. Dell does not provide any warranty as to environment. As we extend our medium beyond the customary print, digital, and Web the accuracy of any information provided through Dell Power Solutions. Opinions expressed in this magazine may not be editions, social media tools such as Twitter provide you with yet another choice on how those of Dell. The information in this publication is subject to change without notice. Any reliance by the end user on the to consume and participate in conversations around the content. We look forward to information contained herein is at the end user’s risk. Dell will not be liable for information in any way, including but not following you. limited to its accuracy or completeness. Dell does not accept responsibility for the advertising content of the magazine or for any claims, actions, or losses arising therefrom. Goods, services, and/or advertisements within this publication other than those of Dell are not endorsed by or in any way connected with Dell Inc.

Printed on recycled paper containing 10 percent post-consumer waste. Please recycle this magazine.

2010 Issue 3

Virtualizing is a great way to cut costs and improve efficiency. But you can’t just deploy the technology and call it good. Is your infrastructure ready for the realities of virtualization?

www.f5.com/dell Feature section Blade solutions for the Efficient Data Center

Game changers

8 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Designed for unparalleled computing density, Dell™ blade servers boost flexibility and minimize costs with innovative power and cooling, intelligent network integration, seamless manageability, and exceptional reliability. How nimble is your IT?

By Robert Bradfield, Chris Christian, and Jeanne Feldkamp

conomic uncertainty, workforce transformation, and shape-shifting technology advances have changed the rules of play for IT decision makers, and data centers are feeling the pressure from all sides. Demand for E computing power will only continue to grow, but a lack of physical space can limit expansion, and construction of new facilities is costly. Many organizations find themselves spending more on power and cooling than they’d budgeted for infrastructure refreshes. Meanwhile, proliferation of disparate platforms, vendor- specific tools, and complex licensing agreements can result in tremendous operational complexity and administrative burden. Building on the core tenets of standardization, simplification, and automation, Dell PowerEdge™ M-Series blade servers and enclosures offer a flexible platform for managing physical, virtual, and logical infrastructures. As a result, blade server environments help eliminate resource silos and optimize data center operations— enabling administrators to allocate resources dynamically and accelerate delivery of strategic IT services. For these reasons and more, blade servers play a pivotal role in fulfilling Dell’s vision for the Efficient Data Center (see Figure 1).

Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. dell.com/powersolutions | 2010 Issue 03 9 Feature section Blade solutions for the Efficient Data Center

High-powered computing keeps its cool • Up to 256 cores per enclosure (4 sockets with IT departments must take a new tack to meet 8 cores per socket and 8 blades per enclosure intensifying demands for computing power while with the PowerEdge M910) increasing reliability and response to variable • Up to 4 TB of memory per enclosure (32 dual workload requirements—often within the confines in-line memory modules [DIMMs] at 16 GB per of tight power and space quotas. Blade servers DIMM with 8 blades per enclosure) help organizations address these challenges in • No extra rack space required for KVM several important ways (see the “Is your data (keyboard, video, mouse) and network switches center a candidate for blade servers?” sidebar in or additional management infrastructure this article). To achieve this degree of computing Efficient design density, Dell blade servers place a premium on Because of their design, blade servers simply efficient processor performance. Low-voltage require less physical space in the data center than Intel® Xeon® processors provide an excellent traditional tower or rack-optimized form factors: combination of high performance and low as many as 16 half-height or 8 full-height blade processor power levels, making them well suited servers can fit into a 10U Dell PowerEdge M1000e for use in high-density blade configurations modular blade enclosure. This efficient, high- running exceptionally demanding workloads. The density design can provide multiple advantages: Intel Xeon processor 5600 series is designed to automatically regulate power consumption and • Up to 60 percent more two-socket density adjust server performance to match application compared with 1U two-socket servers requirements. Intel Intelligent Power Technology • Up to 220 percent more four-socket density dynamically scales power consumption, making compared with 4U four-socket servers power available for critical workloads and

Intelligent Streamlined application data management and workload management

Automated management of data through Rapid delivery of strategic IT services with predictable, scalable, and open data storage: groundbreaking capabilities:

• Minimize storage costs through storage • Automate day-to-day tasks, from workload virtualization, efficient object storage, deployment to image creation automated data tiering, and deduplication • Use best-of-breed technologies as they • Leverage on-demand capacity scaling and develop, without vendor lock-in cloud disaster recovery

Simplified infrastructure management Intelligent infrastructure

Tools for managing and deploying physical, Smart, virtualization-ready infrastructure for virtual, and logical infrastructure: rapid deployment and optimized operations:

• Optimize the management of both physical • Dynamically manage servers, networks, and virtual servers storage, and workloads • Take an incremental, open approach to help • Radically reduce the number of devices to protect IT investments manage, power, and cool—and the tools required to do so

Figure 1. Building blocks for the Efficient Data Center

10 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. conserving power when demand to lend its power and cooling abilities subsides. In addition, Intel Xeon to the server working at 90 percent Is your data center processor models that support Intel capacity. In a modular configuration, a candidate for Turbo Boost Technology provide power and cooling is shared—so those higher-speed execution on demand cooling resources can be directed to blade servers? by using available power to run at a the blade that needs them. higher frequency. Dell’s innovative fan technologies Several key questions help determine when Overall design efficiency help to optimize system temperatures blade servers are a good fit in existing data contributes to lower operating costs and significantly reduce power draw center environments. and requires fewer cooling resources from the cooling infrastructure. The Dell than other blade servers. For example, blade enclosure utilizes three separate What data center workloads are best suited to blade servers? in a Dell competitive power study, and distinct low-impedance airflow Usage scenarios for blade servers are typically 16 Dell PowerEdge M610 blade servers pathways to help ensure ambient air no different from rack or tower servers. Blades inside a Dell PowerEdge M1000e reaches components without being are designed to provide enterprise-class server modular blade enclosure used up to preheated. Power supplies for Dell performance regardless of form factor. 24.1 percent less overall power while blade servers are 80 PLUS platinum idle than 16 HP ProLiant BL460c G6 certified (over 94 percent efficient), the Do economies of scale recommend a servers inside an HP BladeSystem highest rating available. Management minimum number of blade servers per c7000 enclosure. In the same study, features such as Dynamic Power Supply location? 16 Dell PowerEdge M610 blade Engagement further help to optimize Blade server deployments generally become servers used up to 63.6 percent less both performance and efficiency based cost-effective when leveraging seven or more overall power while idle than 14 IBM® on configurable settings. servers across the chassis infrastructure. BladeCenter® HS22 servers inside an IBM BladeCenter H-Series enclosure.1 Intelligent tools work with What about power requirements? existing IT infrastructure Best practices recommend equipping the Innovative power and cooling Dell blade servers are designed to blade server infrastructure with a 208 V power The blade server form factor is designed pair powerful computing capabilities source; 110 V power sources are not efficient to optimize thermal and electrical with optimized connectivity based on enough to support high-density server efficiency. A power and cooling open standards (see the “Dell blade deployments effectively. infrastructure shared among all blades servers up the ante on data center in the enclosure allows resources to be efficiency” sidebar in this article). These How important are provisions for diverted to the servers that need them flexible designs work with existing data networked storage versus local storage? at any given moment. center infrastructures and enable rapid Blade server infrastructures are best served by For example, suppose two blades deployment of advanced technologies. external networked storage such as Internet in the same enclosure are handling Intelligent architecture and streamlined SCSI (iSCSI), Fibre Channel, and network different workloads. One server is systems management capabilities attached storage. External networked storage running at 90 percent capacity, and help simplify and automate IT tasks, enables fast, flexible scalability and high its fans are working at full speed and while energy-efficient infrastructure availability compared with local storage. struggling to keep the server cool. optimizes operations to help reduce The second server is running at only total cost of ownership. What manageability and availability features do blade servers offer? 10 percent capacity and does not need Dell PowerEdge blade servers integrate the full cooling capability of its fans. Integrated networking fabric comprehensive, enterprise-class manageability In a traditional configuration, the Dell modular blade enclosures together with reliability, availability, and underutilized server would have no way integrate edge networking for reduced scalability (RAS) features to help meet rigorous service-level agreements. When this level of 1 Based on SPECpower_ssj2008 benchmark tests performed by the Dell Server Performance Analysis team in June support is not called for, however, rack or tower 2010. Each blade server was configured with two six-core Intel Xeon X5670 processors at 2.93 GHz, 24 GB of Double Data Rate 3 (DDR3) RAM, and two 73 GB, 15,000 rpm Serial Attached SCSI (SAS) hard drives. For complete servers may be a cost-effective alternative. details, visit spec.org.

Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. dell.com/powersolutions | 2010 Issue 03 11 Feature section Blade solutions for the Efficient Data Center

Dell blade servers Dell PowerEdge M710HD blade server This virtualization-optimized blade server offers outstanding I/O up the ante on throughput, maximum memory density with 18 dual in-line memory modules (DIMMs), and robust Intel Xeon processor 5600 series compute data center efficiency power with 12 total cores—all in a half-height form factor with hot- swappable Serial Attached SCSI (SAS) drives or solid-state drives (SSDs). Dell’s latest virtualization-ready blade servers and Designed for enterprises that require a highly dense two-socket enclosures feature intelligent power distribution for form factor with the ability to maximize total RAM, PowerEdge M710HD outstanding performance per watt, balanced system blade servers are well suited for medium and large general-purpose, architecture for efficient memory performance, mission-critical, and memory-intensive applications and databases. innovative power and cooling, integrated networking They may also be a good match for IT departments looking for a highly fabric, and simplified systems management. reliable and redundant half-height configuration that provides both density and I/O flexibility. The PowerEdge M710HD blade server falls between the Dell PowerEdge M610x blade server PowerEdge M710 and PowerEdge M610 blade servers in the portfolio This two-socket blade server based on the because it is designed to offer more memory and more I/O throughput Intel Xeon processor 5600 series leverages than the PowerEdge M610 server and greater density (half height) than the performance and capabilities of the the PowerEdge M710 server. The PowerEdge M710HD blade server PowerEdge M610 blade server in a rack- additionally offers a network daughtercard, which is a flexible version dense, full-height form factor. Coupled with of a conventional LAN on Motherboard (LOM) that allows choices for PCI Express (PCIe) expansion capabilities, integrated connectivity. This modular daughtercard design enables it supports an exceptional range of the integrated network ports to be changed or upgraded as bandwidth connectivity options. requirements evolve over time. The PowerEdge M710HD initially includes Designed for enterprises that need the flexibility and efficiency a quad-port embedded Gigabit Ethernet network interface card (NIC), of blade server technology, but with I/O needs beyond those with two additional network daughtercard options expected to be supported in custom I/O mezzanine cards, PowerEdge M610x blade available soon. These networking options as well servers help enterprises run business applications efficiently and as the expansion options offered by the existing cost-effectively to heighten productivity and innovation. They help I/O mezzanine cards provide a wide range of reduce the data center footprint through consolidation with high- approaches for bandwidth delivery maximization performance virtual machines and expansion-card-based platforms. and customization. Dual embedded The result: enterprises can spend additional time on creating hypervisors and hot-swappable, fault-tolerant business value and reduce time spent on planning, deploying, and hard drives also provide enterprise-class maintaining IT. redundancy for mission-critical applications.

cabling to help simplify deployment and service port aggregators provide the flexibility to scale as well as reduce consumption of expensive their throughput as needed while helping core switch ports. Dell modular blade switches— reduce cabling. SimpleConnect features including the 48-port Dell PowerConnect™ help IT staff avoid complex, time-consuming M6348 blade switch—also help simplify the switch configuration and management tasks. process of upgrading and adding bandwidth. Dell And because the SimpleConnect series of offers these advanced, fully featured integrated networking devices does not require personnel switches as well as a series of port aggregator with specialized skill sets, it also helps to reduce devices known as the SimpleConnect series. For deployment time, training requirements, and those administrators who do not need or desire operating expenses. in-depth management and configuration options Dell FlexAddress Plus features can also for integrated networking, Dell’s SimpleConnect streamline networking management by

12 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Dell PowerEdge M910 blade server Delivering heavy-hitting performance and advanced reliability in a full-height, four-socket blade, PowerEdge Exceptional connectivity M910 servers provide an outstanding platform for Development manager Richard Waldorf highlights the PCIe expansion design and other key deploying large, enterprise-class applications. These Intel capabilities of the PowerEdge M610x blade server. Xeon processor 7500 series–based blade servers feature youtube.com/watch?v=f6iH9LUssaw extensive I/O scalability, pushing the limits of performance with intelligent, expandable processing power that is designed to automatically adapt to diverse, mission-critical workloads. Dell also offers FlexMem Bridge, an innovative, patent-pending technology that enables this leading-edge platform to seamlessly scale memory from 4 GB to 512 GB in either two-socket or four-socket configurations—helping meet changing organizational Maximum memory needs without the need to rip and replace. and high reliability PowerEdge M910 blade servers are well suited for data- Chris Christian outlines the advantages of the PowerEdge M710HD blade server, including demanding applications, private cloud deployments, and heavy the large memory capacity, hot-swappable virtualization environments supporting data center consolidation or drives, and hardware RAID. robust data recovery plans. youtube.com/watch?v=9vH8BYnz_8M

Dell PowerEdge M1000e modular blade enclosure By providing enhanced usability and manageability in a space- efficient form factor, PowerEdge M1000e enclosures are designed for exceptional power efficiency and rapid, straightforward deployment. Efficiency features include active, dynamic Scalable, enterprise-class “zoned” cooling, breakthrough fan technologies, performance ultra-efficient power supplies with dynamic Robert Bradfield runs through the key design features control and power monitoring, and regulators of the PowerEdge M910 blade server and how they support enterprise-class data center deployments. and board design optimized for power and youtube.com/watch?v=JUXNk8jsoY0 cooling efficiency. An LCD panel, wizard-based setup, and centralized configuration through a chassis manager help to streamline deployment. Integrated ship options further help to minimize waste and speed installation.

supporting Media Access Control (MAC) reestablish the relationship between changing LAN or SAN access requirements or World Wide Name (WWN) address the server and its dedicated disk on the or rezoning switches. virtualization without costly proprietary Internet SCSI (iSCSI) or Fibre Channel SAN. hardware requirements. Because it does Dell FlexAddress Plus masks the physical Streamlined management not require separate management tools or MAC and WWN addresses coded to the Dell blade servers are designed with proprietary infrastructure, FlexAddress can hardware so that blade servers can be consolidated management capabilities be extremely cost-effective compared with replaced without breaking the relationship embedded within the hardware, avoiding switch-based alternatives. with the SAN. This approach enables the need for management software to For example, in a typical blade server administrators to re-task or replace aid in deployment and troubleshooting. setup, a MAC or WWN address is coded blade server hardware quickly and Through a single easy-to-use console, to the physical blade server; if the blade easily, without having to carry out the Dell Chassis Management Controller is replaced, an IT administrator must time-consuming processes involved in (CMC) 3.0 helps speed management tasks

Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. dell.com/powersolutions | 2010 Issue 03 13 Feature section Blade solutions for the Efficient Data Center

“Dell blade servers are designed to provide a flexible infrastructure that accelerates delivery of strategic IT services.”

and issue resolution by providing immediate even feature an embedded, dual-media, fail-safe access to system status and alerts. The CMC hypervisor to help protect against downtime in supports automated discovery and rapid server virtualized environments. deployment to help minimize administrative overhead. By automating and simplifying Winning efficiency accelerates delivery hardware configuration and updates for BIOS, of strategic IT services firmware, and drivers in a one-to-many manner, Data centers are grappling with increased CMC 3.0 enables administrators to configure cost and management complexity to maintain multiple blade servers at once.2 existing infrastructure in the face of increasing Virtual File Share features also allow updates performance demands, physical space limitations, and images to be hosted on network storage and and maxed-out power and cooling capacity. made available simultaneously to multiple blades. Blade servers address these challenges in a Robert Bradfield is a senior MAC and WWN addresses can be displayed before cost-effective, energy-efficient form factor that product marketing consultant for blade installation, enabling IT staff to pre-provision enables organizations to shift the balance of IT Dell PowerEdge blade servers with more than 15 years of experience LAN and SAN resources for rapid deployment. spending from fixed maintenance to strategic in the IT industry, including several investments that fuel innovation. years as a systems consultant in Reliability Dell blade servers are designed to provide a the southeastern United States. Reliability is central to all Dell PowerEdge server flexible infrastructure that accelerates delivery of Chris Christian is a senior designs, and the M-Series is no exception. strategic IT services by combining high density product marketing consultant for The PowerEdge M1000e chassis infrastructure with innovative power and cooling, exceptional Dell PowerEdge blade servers with is designed to be 100 percent passive. For manageability, and integrated connectivity for more than 14 years of experience in the IT industry, including example, none of the plugs on the midplane virtually any network, storage, or management enterprise and data center use male connectors, thereby avoiding the environment. Dell PowerEdge M610x, PowerEdge consulting, Web development possibility of bent pins that could necessitate M710HD, and PowerEdge M910 blade servers and design, and application programming and analysis. a chassis overhaul and result in significant together with the PowerEdge M1000e modular downtime. All Dell blade fabrics are designed blade enclosure offer a game-changing array of Jeanne Feldkamp is a to be fully redundant—so if a switch or port features and capabilities designed to optimize data business and technology writer experiences mechanical failure, the blade itself center efficiency—and outplay the competition. based in San Francisco and a frequent contributor to Dell does not lose connectivity. Major subsystem Power Solutions Magazine. components are also redundant and hot pluggable. This redundancy and commitment to Learn more removing and reducing potential single points Dell PowerEdge blade servers: of failure extends to the blade servers as well: dell.com/blades all Dell blade servers offer hot-swappable hard drives and hardware RAID. The PowerEdge The Efficient Enterprise: M710HD and PowerEdge M910 blade servers dell.com/efficiency

2 For more information on CMC 3.0, see “Smooth operator,” by Roger Foreman, in Dell Power Solutions, 2010 Issue 3, dell.com/downloads/global/ power/ps3q10-20100432-foreman.pdf.

14 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved.

Feature section Blade solutions for the Efficient Data Center

Cutting-edge agility

What do a wind farm developer, a clothing manufacturer, a maritime security and tracking services provider, and three institutions of higher learning have in common? They heightened IT efficiency, reliability, and performance using Dell™ blade servers.

fficiency has become the watchword across virtually every line of business. The good news is that technology advances in today’s blade server infrastructures make it a snap to Eoptimize data center efficiency and manageability—enabling organizations around the world to invest newfound IT cost savings in initiatives that inspire innovation and accelerate growth. The success stories highlighted here represent a small but diverse sample of how organizations are putting Dell PowerEdge™ blade servers to work in ways that help maximize IT efficiency and flexibility, whether to consolidate data center resources, boost savings on power and cooling, virtualize servers for rapid deployment, or process data-intensive applications on a high- performance computing (HPC) platform. Their results may surprise you.

16 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. RES: Running like the wind on understand its needs, and recommended Dell blade servers an energy-efficient HPC cluster built on For rapidly growing wind farm developer 16 Dell PowerEdge M610 blade servers with Renewable Energy Systems (RES), a power-efficient Intel® Xeon® X5550 processors running HPC platform is mission critical. Having built Platform Cluster Manager – Dell Edition software. more than 80 wind farms around the globe, RES Mounted in a PowerEdge M1000e modular continues to expand into new regions, including blade enclosure and connected with Dell Turkey and South Africa, and had to upgrade its PowerConnect™ M6220 switches, the cluster existing Dell HPC solution to quickly and accurately easily fits in the space available, and includes Powerful assess these complex emerging markets. a Dell PowerVault™ MD1000 direct attach array performance

Minimizing its carbon footprint and saving for high-volume storage. In this video, learn more space were key requirements, particularly because “Now we can complete the job in two to about how Dell blade servers an RES sister company was hosting the system three weeks—approximately 20 times faster— helped RES meet its need for a high-performance, until a new data center could be built. Because which means we can get wind farms in place energy-efficient cluster. IT resources were scarce, simplified systems quicker,” says Peter Stuart, technical manager at management was also essential. RES. In addition, the company has reduced power youtube.com/watch?v= cwkDh-_pVnk The Dell Global Infrastructure Consulting consumption by approximately 75 percent and Services team worked closely with RES to freed up valuable IT staff time.

Cornell University: Simplifying platform and DataDirect Networks storage. HPC solutions on the TeraGrid Force10 Networks switches were sourced HPC is the backbone of the work performed through Dell. This HPC cluster allows a wide by the Cornell University Center for Advanced range of scientific researchers and students to Computing (CAC), which supports Cornell use MATLAB to help reduce the time to solution researchers and students from scientific in a seamless experience, without having to disciplines across the United States. The tackle the complexity and intricacies of parallel center recently received a National Science computing. For example, external researchers Foundation grant to deploy the MATLAB were able to remotely access all 512 cores of the numerical computing environment to a cluster MATLAB cluster to model a hepatitis C virus, a called “MATLAB on the TeraGrid.” The goal was major cause of liver disease worldwide. to demonstrate a working model for high- “We’re confident that scientific users and performance utility computing with MATLAB, engineers from across the nation will benefit from and to encourage other software vendors to the reliability and performance of the Dell blade develop similar capabilities. solution while, on the IT side, Cornell will benefit The cluster uses Dell PowerEdge M600 blade from reduced power and space requirements and servers with Intel Xeon 5420 processors running the ability to maintain the blades with in-house the Microsoft® Windows® HPC Server 2008 staff,” says CAC director David Lifka.

“We’re confident that scientific users and engineers from across the nation will benefit from the reliability and performance of the Dell blade solution.” —David Lifka Director at Cornell CAC Dell blade servers optimize data center April 2010 efficiency—and boost organizational agility— through exceptional computing density, energy efficiency, network integration, and manageability

Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. dell.com/powersolutions | 2010 Issue 03 17 Feature section Blade solutions for the Efficient Data Center

Pacific Sunwear: Taking a shine on Dell PowerEdge M610 blade servers with to huge time savings Intel Xeon 5500 series processors and a Dell Performance and reliability are essential for EqualLogic™ PS Series iSCSI (Internet SCSI) clothing chain Pacific Sunwear (PacSun) of storage area network (SAN). California. This top teen fashion retailer relies on “From a three-year ROI standpoint, we its high-speed reporting system to collect data could easily double our return with a virtualized from stores daily, spot trends, and coordinate environment versus just replacing old servers with deliveries with its distribution center in Kansas. new physical machines,” says Ira Ham, director This system helps ensure that the latest products of network and information security at PacSun. reach stores in time to meet demand. If an IT “Virtualization could also help us deploy new system goes down, it can affect the entire sales applications faster and open up space in the data chain and result in missed opportunities. center for growth.” When PacSun needed to refresh its PacSun has also dramatically decreased its infrastructure to increase availability, reduce costs, operating expenses by eliminating more than simplify management, and enhance business 100 servers through virtualization, while simplified agility, it created a virtualized data center with storage management has resulted in a 20 percent VMware® vSphere™ virtualization software running reduction in administrative time.

Thomas College: Making the In the data center, Thomas is consolidating on grade in server reduction virtualized Dell PowerEdge M600 and PowerEdge Server consolidation through virtualization is M610 blade servers with Intel Xeon processors important to private liberal arts school Thomas running the Microsoft Hyper-V™ platform—a College in Maine, which serves more than move that has helped the college consolidate 1,100 students with a cutting-edge IT infrastructure its footprint by 50 percent and reduce heat designed to offer the latest Microsoft communication generation by more than 30 percent, thereby and collaboration tools. The college’s IT department saving on power and cooling costs. Microsoft needed a simplified system that would allow it to System Center Virtual Machine Manager enables provide high-performance laptops, desktops, and IT staff to repair and upgrade physical hosts portable computers while increasing management without interrupting users simply by moving the efficiency in the data center. The college became virtual machine images to a different blade server, an early adopter of Microsoft Windows 7 on eliminating half of the planned downtime. And IT Dell OptiPlex™ desktops and Dell Latitude™ laptops, staff can also set up a virtual machine in half an enabling IT staff to image computers more quickly hour, making it easy to accommodate evolving than in previous OS versions. needs of the Thomas faculty.

RES Cornell University Pacific Sunwear Thomas College 75% 512 cores 100 servers 50%

An energy-efficient cluster Cornell CAC demonstrated Virtualization on Dell blade Consolidation on Dell of Dell blade servers a utility that helped simplify servers helped Pacific blade servers enabled helped RES reduce power a complex scientific Sunwear eliminate more Thomas College to reduce consumption by roughly solution involving access than 100 physical servers its server footprint by 75 percent while minimizing to all 512 cores of a Dell and reduce administrative 50 percent and lower heat the server footprint. blade server cluster. time by 20 percent. generation by 30 percent.

18 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Pole Star: Navigating maritime as a flexible and easy-to-deploy solution. Working security requirements with Dell, Pole Star IT staff designed and deployed When the International Maritime Organization VMware vSphere virtualization on Dell PowerEdge issued a regulation stating that every ship at sea M600 blade servers with Intel Xeon E5410 must report its position four times a day, Pole Star processors and Dell EqualLogic storage, protected Space Applications Ltd., a UK-based provider of by Dell ProSupport for IT. maritime security and satellite-enabled tracking The move to Dell blade servers cut the technology, anticipated a massive increase in sales physical server footprint by 83 percent, and volume. With the new law pending, Pole Star had Pole Star estimates that the current virtualized less than 12 months to prepare its infrastructure environment will serve business needs for the next for the rise in demand. five years even with an anticipated 50 percent The organization was running Dell servers in growth in business volume within the next three hot standby clusters for high availability, but this years. “We’re expanding rapidly, and having the approach was relatively expensive and required new technology to support that process is incredibly hardware every three years. The company needed a important,” says James Bayley, consultant head of powerful environment that could scale quickly with IT for Pole Star. “The Dell solution has delivered on minimum expense, and recognized virtualization the promise of business agility.”

Stony Brook University: Leading the cluster of five PowerEdge M905 blade servers with pack with Dell PowerEdge servers 20 six-core AMD Opteron™ 2435 processors and Dell performance, reliability, and speed were a 40 Gbps InfiniBand interconnect provided by all put to the test—literally—when the Stony Mellanox Technologies. Brook University team at the Supercomputing As a result of Dell’s rapid response and the Conference (SC) Student Cluster Competition system’s ease of deployment and simplified needed the components of an HPC cluster to beat configuration, cluster setup was quick, and the competing universities in processing large data students had the time they needed to prepare for sets and compute-intensive scientific applications. the event. The Stony Brook team came away with In the spirit of preparing students and the honors, and Dell performance was critical to attracting interest in HPC, SC sponsors an annual success: “The amount of memory we had on the competition in which students compete on the Dell machines allowed us to run much larger data basis of benchmarking and scientific problem sets than the other teams, which was definitely a solving. A quick response from Dell enabled the factor in our winning,” says Aaron Pellman-Isaacs, Stony Brook University team to build an HPC senior biology major and team leader.

Dell PowerEdge blade servers: Pole Star Stony Brook University Standardized platforms for flexible growth As IT decision makers run a gauntlet of financial and organizational 83% 1st place challenges, Dell PowerEdge blade servers ease the way with a compact form factor designed to increase operational efficiency, reliability, Virtualization on Dell blade Stony Brook University and manageability while saving big on power and cooling expense. servers cut Pole Star’s finished first at the SC PowerEdge blade servers also provide diverse IT environments with physical server footprint by Student Cluster Competition the computational density and intelligent fabric integration needed 83 percent while supporting thanks to Dell blade server to optimize availability and performance for a world of applications. an anticipated 50 percent performance. The six success stories featured here demonstrate that big results growth. can indeed come in a very small package.

Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. dell.com/powersolutions | 2010 Issue 03 19 Feature section Blade solutions for the Efficient Data Center

Smooth operator

By Roger Foreman

Built by implementing an innovative, user-driven development process, the Dell™ Chassis Management Controller (CMC) 3.0 embedded management tool enables administrators to simply and intuitively manage multiple blade servers from a single console.

he Dell Chassis Management development team wanted to dramatically enhance Controller (CMC) is a management the simplicity and usability of the management tool embedded in Dell PowerEdge™ tool—with the goal of enabling busy system T M1000e modular blade enclosures administrators to spend as little time as possible that provides powerful functionality to help managing servers, leaving them with additional administrators simply and easily manage blade time for other, increasingly complex tasks. servers and blade server enclosures. Functionality The development team utilized an CMC 3.0 available in the CMC includes dynamic power innovative, iterative, user-driven design process interface video management, blade server component to help streamline the management tools for monitoring, and real-time reporting of power administrators of PowerEdge blade servers This video provides a brief tour of the GUI in the consumption, temperature, and other indices of and modular blade enclosures. The team CMC 3.0 management tool blade server and chassis health. collaborated with the Dell Experience Design embedded in PowerEdge M1000e enclosures. To help make chassis management even Group to begin designing an interface that easier and more intuitive than using traditional would be robust yet intuitive and easy to use, delltechcenter.com/ management tools, Dell substantially redesigned and then built and tested the software using the video/10037711/cmc+3.0 the CMC interface for version 3.0. In particular, the Agile programming methodology, an iterative

Administrators can access the CMC 3.0 interface through a standard Web browser, and monitor up to 16 blades on a single screen.

Browser title development method that incorporates with chassis blade name user feedback continuously throughout the development process. The result: CMC 3.0 offers a rich set of Tree pane features and a graphical user interface (GUI) with status that incorporates the Dell Clarity E style, an approach that provides dynamic, self- Updated refreshing pages; an updated color scheme; color scheme with black and an intuitive home page that allows background administrators to perform most monitoring and management functions from a single Enhanced screen, and with significantly fewer mouse graphics clicks than in previous CMC versions.

Collaborative design Figure 1. Home page for Dell PowerEdge M1000e chassis status in CMC 3.0 A key goal for the CMC 3.0 development team was to design a tool that provides robust functionality but is also intuitive. To help achieve that goal, the team enlisted the help of the Dell Experience Design Group, a cross-functional team of design experts at Clickable graphic to provide Dell that comprises UI designers, usability detailed views engineers, and Ph.D.-level cognitive scientists of specific components and psychologists. In collaboration with this group, the CMC 3.0 development team first used extensive Links for direct access to input from target users to identify and frequent actions carefully catalog typical chassis management tasks that administrators perform. Identified tasks included managing power usage, monitoring blade server components, Figure 2. Health, power consumption, and temperature status for a blade and quickly consulting indices of server server in CMC 3.0 health such as temperature and power consumption; additional tasks included adding users, configuring servers, and diagnosing and resolving errors. The team then used this feedback to Chassis ambient design a UI prototype based on the Dell temperature Clarity E style, which is now the standard for (input air) Dell OpenManage™ systems management Current blade tools. The style employs a wide range of temperature dynamic, self-refreshing screens, colors, and features, and offers a consistent look and feel across systems management tools in the Dell

OpenManage suite. Overall server Using the Agile programming methodology, health a Dell team of more than 20 programmers developed incremental features during Figure 3. Sensors for ambient chassis and blade server temperatures in CMC 3.0

Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. dell.com/powersolutions | 2010 Issue 03 21 Feature section Blade solutions for the Efficient Data Center

“The heart of CMC 3.0 is its intuitive home page, which enables administrators to use a single Web-style page to perform most monitoring and management tasks.”

every three-week sprint. Sprints concluded with host name as defined by the OS running on a live demonstration to solicit feedback and the blade, as well as system-wide temperature final approval that each feature met the need to monitoring, which enables administrators to optimize simplicity, functionality, and ease of use. monitor not only chassis ambient temperature, but also individual blade temperatures (see Intuitive management Figure 3). CMC 3.0 also expands Lightweight CMC 3.0 offers a redundant monitoring and Directory Access Protocol (LDAP) functionality Systems management tool that is embedded in Dell to include support for a range of LDAP services management for PowerEdge M1000e modular blade enclosures. such as Microsoft® Active Directory® directory Dell blade servers The CMC 3.0 user interface, accessible through services, Novell® eDirectory™ directory services, standard Web browsers, is automatically and open source directory services such as In this video, Roger Foreman discusses the included with the hardware and does not OpenDS and OpenLDAP. benefits of using CMC 3.0 require special installation. to help streamline systems maintenance tasks. The heart of CMC 3.0 is its intuitive home User-driven expansion page, which enables administrators to use a single By combining enhanced graphics and usability youtube.com/watch?v= Web-style page to perform most monitoring and features with extensive user feedback, the s-ASdMtJY5k management tasks (see Figure 1). Using the home user-driven design of CMC 3.0 helps deliver page, administrators can see the status of chassis robust, streamlined functionality through an components at a glance and perform key functions intuitive, easy-to-use interface. Building on such as power cycling, power management across the design process of CMC 3.0, Dell plans to blades, and error diagnosis. The home page is incorporate the same look and feel and design also dynamically updated, enabling administrators methodology across the Dell OpenManage to access up-to-date status and error information systems management suite to help simplify without having to manually refresh the screen. management, streamline operations, and CMC 3.0 is also scalable: administrators can see enhance administrator productivity. and monitor up to 16 blades on a single screen, and

can easily drill down into individual components Roger Foreman is a senior product manager to view detailed information. For example, an on the Dell OpenManage Marketing team administrator can click a power supply to quickly with responsibilities for CMC firmware and management of blade servers. Previously, he led see its properties, its status, and links to other the team that created delltechcenter.com. actions or detailed information. Similarly, clicking an individual blade server brings up key status Learn more indicators such as health, power consumption, and temperature, and also provides server properties Dell systems management: and links to other actions (see Figure 2). dell.com/openmanage Other features of CMC 3.0 include automatically populating the slot name that Dell PowerEdge blade servers: appears in the browser title or tab with the dell.com/blades

CONTROL THE CLOUD

CM Public. Private. Hybrid. Virtual. Physical.

MY No matter which approach you choose, CY use BMC So ware to take full control CMY of your IT operations from a single, K uniﬁed platform.

Learn more at www.bmc.com/controlthecloud

Cost-effective tiering for virtualized iSCSI SANs

By Achmad Chadran and Keith Swindell

Tiering in a Dell™ EqualLogic™ PS Series SAN helps organizations boost resource utilization and enhance service levels by deploying storage on the most cost-effective tiers based on performance, accessibility, and criticality requirements.

ampant data growth and stringent may be performance-constrained while others data retention policies are placing use storage that is much more costly than they huge demands on enterprise storage require. IT organizations can manage these Rresources. Several factors have challenges by classifying data based on specific contributed to this explosion of data, including criteria and then assigning it to an appropriate increasing application diversity and specialization, storage tier, helping increase resource utilization, the proliferation of automated data collection enhance service levels, and reduce costs. tools for advanced decision support and other processes, and regulatory standards for Deploying data on storage tiers data retention. Storage tiering has emerged as a strategy for To manage this increase in demand, many balancing performance and cost-efficiency. In a organizations have consolidated data on storage tiered storage deployment, data can be moved area networks (SANs) to help dramatically simplify to different storage media based on specific storage management and increase utilization. requirements. Application data with extremely When using SANs as part of a virtualization strategy high performance requirements or high access to consolidate resources, balancing performance rates, for example, may be deployed on high- and cost-efficiency can be challenging, because performance storage, while data with low simple SAN consolidations limit the flexibility to performance requirements or access rates may Enhancing virtual optimize storage for multiple applications. be deployed on lower-cost storage. In addition desktop scalability A typical SAN, for example, might utilize to enhancing cost-efficiency, storage tiering also Automated tiering and load 10,000 rpm Serial Attached SCSI (SAS) drives helps ensure that performance-optimized arrays balancing of data across throughout the deployment as a simple are available to meet service-level agreements solid-state and spinning media enhances scalability approach to data consolidation. Different (SLAs) and other criteria. for virtual desktops. In this applications within that deployment, however, Implementing a storage tiering strategy can be video excerpt of his keynote at Citrix Synergy 2010, Michael may have very different performance labor intensive, however. It requires administrators Dell speaks about virtualized requirements: some applications may produce to predict which storage workloads will be iSCSI SAN capabilities that can drive high levels of efficiency data with a heavy I/Os per second (IOPS) load, associated with which applications, define tiering and performance for virtual which cannot be met easily using 10,000 rpm requirements based on application mix, commit desktops. media, while other data—such as backups, arrays at each tier to accommodate current equallogic.com/resourcecenter/ archives, and file data such as documents and requirements plus growth, and continuously video.aspx?id=9493 media—may be rarely accessed. As a result, monitor workloads to validate the tiering when using a single-tier SAN, some applications architecture and adjust as necessary.

Member 1 Member 2 Member 3 Member 4 (RAID-10) (RAID-5) (RAID-5) (RAID-50)

Web server volume Regulatory data volume 2

Regulatory data volume 1 E-mail volume

Default storage pool

Dell EqualLogic PS Series group

Figure 1. Multiple members of a Dell EqualLogic group in a single storage pool

To help organizations take advantage of streamlined management, and efficient data the performance and cost-efficiency benefits protection and recovery. of tiering, Dell EqualLogic PS Series Internet With virtualized storage as a central design SCSI (iSCSI) SAN arrays provide a rich set of tenet, each EqualLogic SAN comprises one or more tools to tier data based on multiple criteria. In EqualLogic arrays (members) connected to an IP particular, the EqualLogic solution offers three network and managed as a single system. Individual different techniques for storage tiering: automatic arrays are grouped into one or more virtualized performance load balancing within an EqualLogic storage pools—essentially, a partitioned “SAN storage pool, RAID preference configuration within a SAN”—which may be homogeneous, with tiering within an EqualLogic storage pool, members utilizing the same type of storage media and pool-based tiering using multiple storage and RAID level, or heterogeneous, with members pools within an EqualLogic group. EqualLogic utilizing different storage media and/or PS6000XVS and EqualLogic PS6010XVS arrays can RAID levels. A volume, or a logical unit (LUN), is Why tier storage? extend the value of tiering by enabling tiers within assigned to a storage pool (see Figure 1); data on Although SAN-based each volume (see the “Sub-volume storage tiering that volume is striped across all the disks in that pool storage consolidation can help manage spiraling in Dell EqualLogic arrays” sidebar in this article). that contain part of the volume, per the RAID policy. enterprise data, application Automated storage tiering on EqualLogic SANs EqualLogic SANs support automated storage diversity can pose challenges helps reduce costs, enhance efficiency, and boost tiering and can be monitored through EqualLogic for maintaining both performance and efficiency. the performance of storage infrastructures. SAN HeadQuarters (SAN HQ), a Microsoft® This video makes the case Windows® OS–based application with an easy- for SAN tiering and offers a peek at the advanced Tiering storage on Dell EqualLogic to-use interface that offers visibility into storage tiering capabilities offered on PS Series arrays IOPS and utilization at the volume, pool, member, EqualLogic PS Series SANs.

EqualLogic PS Series iSCSI SANs provide a and group levels. Organizations can deploy the equallogic.com/whytier scalable, high-performance, easy-to-manage combination of automatic performance load platform with which organizations can simply and balancing, RAID preference configuration tiering, cost-effectively consolidate enterprise storage. and pool-based tiering techniques that best suits The arrays offer tremendous capacity and utilize their needs, and can easily modify the storage a virtualized storage architecture that enables tiering strategy as needed. When using EqualLogic storage to be scaled rapidly and efficiently firmware version 5.0, SAN HQ is also able to without reducing performance or disrupting provide information about the amount of data that availability. EqualLogic SANs incorporate a range is in one of three I/O categories—high, medium, of features designed to support rapid installation, or low—to further assist the administrator in

understanding the overall needs of the storage volume to be associated with a Pool-based tiering organization’s data and usage patterns. RAID preference. This option overrides the EqualLogic SANs also offer pool-based automatic performance load balancing. tiering, which is designed to deliver Automatic performance load balancing A volume expected to have a very high maximum control over storage tiering. In Automatic performance load balancing IOPS requirement might be associated an environment using this technique, each is the default storage tiering policy on with a high-performance RAID policy such storage pool is a tier, and volumes are EqualLogic SANs. When using automatic as RAID-10, while a volume expected assigned to individual pools. For example, performance load balancing, the arrays to have a low IOPS requirement might an organization might choose to deploy automatically tier data within a pool based be associated with a low-performance two storage pools, or two tiers—one for on random IOPS for a volume, with no RAID policy such as RAID-50. mission-critical data and one for bulk administrator intervention required. The EqualLogic firmware stores storage data that is not accessed as heavily This policy utilizes algorithms built into the volume data on an array or set of arrays (see Figure 2). The storage pool for the EqualLogic firmware to monitor IOPS at the that matches the preferred RAID policy mission-critical application data can contain volume level. When appropriate, the firmware level for that volume if it is available high-performance arrays and use a RAID can move a volume to arrays within the pool within the pool—that is, a volume with level consistent with the read/write patterns that are configured for RAID-10, which is a RAID-10 preference is stored on a associated with the applications accessing suitable for high levels of random IOPS. RAID-10 array, if available. If an array that pool. The storage pool for the bulk data Automatic performance load balancing using the preferred RAID policy is not can contain high-capacity arrays configured is designed to be an excellent all-purpose available, an alternative array or set of to maximize storage capacity and minimize tiering mechanism, and is especially well arrays would be used. Although storage cost for long-term storage. suited for larger enterprises that may tiering based on RAID policy preference When adding arrays to the SAN, not have had the opportunity to analyze is automatic, it is not guaranteed: the administrators can deploy them in application storage workloads and that firmware attempts to satisfy the preference, the appropriate pool, and they can have four or more arrays in a single pool. but if a RAID level is over-allocated, configure the pools—and volumes Because this policy evaluates usage patterns some volumes will be moved to other assigned to them—based on a wide and moves data very gradually, it is well RAID levels to maintain appropriate range of criteria, including application suited for organizations with large amounts free space. Additionally, the pool must performance requirements, geographic of data or data that does not need to be have arrays of different RAID types for location, business units, budgetary rapidly migrated from one tier to another. this tiering technique to be effective. considerations, and more. One RAID preference configuration tiering example strategy would be to divide RAID preference configuration tiering is well suited for organizations that need storage into the following four tiers: Another storage tiering technique available to control disk resources, maintain good with EqualLogic SANs is RAID preference performance for tier 1 applications, and • Tier 0: Utilizes SSDs for applications that configuration tiering, which allows a support application-by-application SLAs. have a high number of random reads

Switched Gigabit Ethernet or 10 Gigabit Ethernet network

Figure 2. Separate Dell EqualLogic Member 1 Member 2 Member 3 Member 4 storage pools configured for specific data requirements Production applications Backup and archive volumes volumes

Application storage pool Default storage pool

Dell EqualLogic PS Series group

Storage tiering—matching storage media to specific version 5.0, which extends existing auto-tiering algorithms to organization requirements—has emerged as a way to help enable tiering within volumes on these models. optimize storage performance while minimizing costs. Most Sub-volume storage tiering can be especially useful storage tiering occurs at the volume level; that is, a volume for volumes associated with virtual desktop infrastructure is assigned a storage tier based on criteria defined by the deployments. Each virtual desktop instance is typically deployed application, such as performance, capacity, and availability. This with two parts: a high-read-intensity “hot” master image, and a approach treats all data on that volume as equivalent, when in less-heavily accessed “warm” desktop that stores data unique fact some of the data may be very active, while the majority of to that virtual machine. With sub-volume tiering, organizations the volume may be relatively inactive. can leverage extremely high-performance solid-state drives EqualLogic PS6000XVS and EqualLogic PS6010XVS arrays (SSDs) to service the hot master images, and use 15,000 rpm from Dell, however, offer storage tiering at the sub-volume Serial Attached SCSI (SAS) drives for “warm” data for each virtual level, enabling discrete workloads to be isolated and tiered machine. This type of tiering enables organizations to leverage separately within a volume. Each array incorporates both solid- the tremendous performance advantages of SSD-based storage state and spinning media and includes EqualLogic firmware without having to provision the entire volume with SSDs.

along with low latency and modest capacity the EqualLogic SAN HQ monitoring tool to requirements observe storage usage patterns through an easy- • Tier 1: Utilizes 15,000 rpm SAS drives in a RAID- to-use console that provides not only visibility Achmad Chadran is a 10 configuration for I/O-intensive applications into storage IOPS and utilization, but also a range storage solution marketing manager in the Dell Large • Tier 2: Utilizes 10,000 rpm SAS drives in a RAID- of near-real-time and trend data statistics for Enterprise Business Unit. 50 configuration for moderate-I/O applications monitoring tiering schemes. Dell Services also • Tier 3: Utilizes SATA drives in a RAID-50 provides a broad range of storage optimization Keith Swindell is a product configuration for minimally accessed data services that can include tiering planning, manager on the EqualLogic Engineering team at implementation, and management. Dell with over 20 years’ Many other possibilities exist with pool-based By enabling organizations to cost-effectively experience in designing and storage tiering, and the appropriate combination tier storage based on specific requirements, with implementing effective IT solutions to solve complex of disk types and RAID types chosen will depend both administrator-defined and automatic storage business problems. on the needs of the organization. Pool-based tiering, an EqualLogic PS Series iSCSI SAN from storage tiering can be well suited for organizations Dell helps enhance both efficiency and service with easily defined storage requirements and a levels. And by enabling storage tiering at the need for maximum control over data placement. sub-volume level, EqualLogic PS6000XVS and EqualLogic PS6010XVS arrays enable organizations Enhancing efficiency with flexible, to deploy next-generation automatic storage cost-effective storage tiering tiering technology to help maximize flexibility and Organizations can select an appropriate storage minimize administration. tiering strategy by considering criteria that are the most relevant to their needs; these Learn more criteria can include metrics such as access Dell EqualLogic PS Series: frequency, performance, mission criticality, dell.com/psseries and data protection, or can be based on data dell.com/equallogic type. Analyzing current and past storage usage also helps organizations identify an appropriate Dell EqualLogic storage tiering: storage tiering strategy. Administrators can use equallogic.com/tiering

anaging a database environment presents common challenges across midsize and large organizations, Mincluding controlling costs, optimizing resource utilization, scaling the environment effectively, and isolating resources to help ensure maximum availability and performance. Many organizations are also looking to reduce power, cooling, and space requirements in the data center. At the same time, organizations may lack sufficient IT resources to deploy, manage, and maintain complex or customized environments. Combining Oracle Database 11g and multi- protocol Dell/EMC unified storage, along with Oracle Direct Network File System (DNFS) included in Oracle Database 11g, helps organizations efficiently address these challenges. Oracle DNFS offers a highly tuned protocol that helps Dell/EMC unified storage deliver high performance, low server processor overhead, and simplified administration. Together with virtualization, the Dell/EMC unified storage platform enables organizations running Oracle database software to maximize performance and scalability based on workload requirements while helping simplify backup operations, accelerate deployments with cloning, and increase the Streamlining performance of disaster recovery operations. enterprise database Capitalizing on the advantages of NAS and SAN protocols Organizations deploying Dell/EMC unified storage environments can migrate Oracle databases between network attached storage (NAS) protocols and storage area for business continuity network (SAN) protocols without downtime, enabling administrators to match database operations to By Annette Cormier, Jason Kotsaftis, and Jeff Browning the appropriate protocol and infrastructure. For example, database replications used in testing and development as well as backup and recovery The combination of Oracle® Database 11g, Oracle processes can be seamlessly migrated to economical, Direct Network File System, and multi-protocol high-availability Network File System (NFS)–mounted storage. NFS is a broadly adopted and cost-efficient Dell/EMC unified storage offers a single platform protocol, and does not require specialized training to help unlock the power of virtualization and to use. Simultaneously, production databases requiring maximum availability and low latency can optimize database performance. be maintained on high-performance Fibre Channel Protocol (FCP) storage. Enterprise flash drives (also known as solid-state drives) can also enhance performance in many database use cases.

28 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Dell/EMC unified storage enables this 500 combined storage model by consolidating NAS 450 OS-bonded KNFS 400 Oracle DNFS and SAN functionality and performance into a 350 single shared storage system that supports both 300 NFS and Fibre Channel. Organizations deploying 250 Oracle Database software can choose either 200 standard kernel NFS (KNFS) or Oracle DNFS 150 Throughput (MB/sec) Throughput with Dell/EMC unified storage. DNFS provides 100 additional benefits and enhancements beyond 50 what KNFS offers. 0 One port Two ports Three ports Four ports

Overcoming NFS I/O performance Figure 1. Comparing data throughput for standard KNFS and Oracle DNFS using a DSS workload variability Oracle DNFS embeds the NFS client directly into the Oracle Database 11g database resulting in I/O performance variability. Oracle application kernel. Bypassing the OS for DNFS is designed to overcome this variability in concurrent I/O helps reduce memory and multiple ways: processor utilization, thereby enhancing performance. DNFS is optimized for database • OS independence: Hosting the NFS client workloads and supports asynchronous I/O, directly in the database kernel instead of in the which can make this protocol preferable to OS kernel helps ensure consistent performance Innovative KNFS in some organizations; DNFS can also across operating systems. storage load balance more efficiently than OS-based • Asynchronous direct I/O: The DNFS kernel efficiencies KNFS clients that are not optimized. enables asynchronous direct I/O, a highly This series of on-demand In validation testing of Oracle DNFS using efficient approach that allows database I/O to Webcasts offers best a decision support system (DSS) workload, continue while other requests are submitted practices for tiered storage DNFS showed significantly increased maximum and processed. DNFS also uses database deployments in Oracle environments, performance throughput compared with KNFS (see Figure 1).1 caching with asynchronous direct I/O to help benchmarks with enterprise This increased performance benefits ensure almost immediate data writes, which flash drives for ultra-high performance, and using administrators and users of highly helps reduce integrity risks. advanced features of demanding business intelligence and DSS • Efficient setup: The DNFS client manages Oracle Database 11g.

applications, helping to maintain high levels load balancing and high availability to help dellenterprise.com/goto/ of responsiveness for these mission-critical simplify network setup, which helps reduce emcoracle capabilities. dependence on network administrators and A key performance challenge in database avoids the need to set up network subnets and environments is that IP storage offloads the bond ports such as Link Aggregation Control file and volume management operations to Protocol (LACP) bonding. the storage array, but works with the host OS • OS write locking: DNFS overcomes OS write in caching I/O read/write operations to the locking, which can be inadequate in some database. In non-database environments, the operating systems and cause I/O performance use of the OS to mediate cache and control bottlenecks in others. DNFS also helps reduce the I/O for the NFS protocol is standard. In database server processor and memory database environments, however, the read/write usage by avoiding the overhead of copying and asynchronous nature of database I/O is data between the OS memory cache and the typically not managed well by OS client kernels, database System Global Area (SGA) cache.

1 Based on EMC testing performed in October and November 2008 using an Oracle Real Application Clusters (RAC) architecture consisting of four Dell™ PowerEdge™ 2950 server nodes configured with dual-core processors at 3 GHz, 16 GB of RAM, and the 64-bit Microsoft® Windows Server® 2003 OS, connected to an EMC Celerra® NS-40 storage array.

ASM DATA

ASM LOG1

ASM LOG2

ASM SNAPDATA Dell/EMC CX4 production array ASM SNAPLOG1 (EMC SnapView snapshot) ASM SNAPLOG2

NFS FLASH

Source Oracle Target Oracle NFS ARCH Figure 2. Migrating an online Oracle RAC cluster RAC cluster database from a SAN to NAS (Oracle Data Guard NFS CRS and RMAN) NFS TGTDATA

NFS TGTLOG1 Dell/EMC NS target array NFS TGTLOG2 Virtualized server IP storage network FCP storage network 1 FCP storage network 2 Oracle Real Application Clusters (RAC) interconnect 1 Virtual migration server Virtual EMC Replication Oracle RAC interconnect 2 (Oracle Data Guard Manager server Back-end network and RMAN) (backup)

Migrating a virtualized database a consistent backup using capabilities As an alternative to mounting the RMAN from a SAN to NAS provided by the Dell/EMC NS-480 array. image backup directly, administrators By consolidating SAN and NAS functionality Administrators can perform the backup can optionally use Oracle Data Guard to while supporting both Fibre Channel and using the following steps: place the target database into continuous NFS, Dell/EMC unified storage helps simplify recovery mode. Log ship and log apply database migration. The Dell/EMC NS-480 1. Use EMC® Replication Manager software processes then catch up the target to array, for example, enables organizations to automate application-consistent the production version, and Data Guard to efficiently move a virtualized Oracle snapshots. failover can be used to retarget the NFS database from a SAN to NAS-mounted 2. Use EMC SnapView™ software to create a database. With either method, the result is storage and across protocols (see Figure 2). snapshot of the database. that a production Fibre Channel–mounted Using this capability, organizations 3. Mount this backup and the NFS target database can be migrated to NFS with little can easily support testing, backup, and array on the migration server, which may performance impact and no downtime. recovery on NFS storage and production on be either a VMware® virtual machine or a Administrators can migrate an online Fibre Channel storage from within a single physically booted server. Oracle database from NAS to a SAN storage system. Administrators can also 4. Create an Oracle Recovery Manager with the same sequence, but using EMC deploy storage replication and continuous (RMAN) backup of this database as SnapSure™ software instead of SnapView data protection on the same system using a a database image so that the data and using Fibre Channel as the target different FlexAddress I/O port than the one files are written directly to the target instead of NFS. This nondisruptive migration used by the production database.2 NFS mount. can be extremely beneficial when a Migrating an online Oracle database 5. Switch the migration database to the successful business grows and needs to from a SAN to NAS starts with performing copy on the NFS target. scale its IT infrastructure from NAS to a

2 For more information on replication in Oracle database environments, see “Maximizing replication efficiency in Oracle environments,” by Annette Cormier and Jason Kotsaftis, in Dell Power Solutions, 2010 Issue 3, dell.com/downloads/global/power/ps3q10-20100260-cormierorcl.pdf.

30 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. SAN. This may be the case when a database copied, so only space for the modified data is workload grows enough that it requires the required. SnapSure reads the original blocks low-latency performance and scalability that a and the unchanged blocks remaining in the PFS Fibre Channel SAN provides, such as a highly according to a bitmap and blockmap data- transactional online transaction processing (OLTP) tracking structure. These blocks combine to infrastructure. The ability to nondisruptively provide a complete point-in-time file system migrate between NAS and a SAN is one primary image. The snapshot reads blocks A and B from Annette Cormier is reason that an organization might choose a multi- the source logical unit (LUN) and block C from a solutions marketing manager for Dell/ protocol storage system. the reserved LUN. EMC with 20 years of experience developing Performing high-speed backups Supporting a highly efficient and bringing to market enterprise storage, and database cloning architecture network management, Dell/EMC unified storage also enables high-speed, Server processor capacity is one of the and security products. nondisruptive array-based snapshots, which can most precious resources in enterprise IT Jason Kotsaftis is be used for either quick data restoration or as the environments, particularly in database a director in EMC basis for full database backups. EMC SnapSure architectures. Oracle Database 11g and Oracle Technology Alliances with snapshots can be checked into the Oracle RMAN DNFS support highly tuned NFS systems that 14 years of experience repository, and are integrated with RMAN for Dell/EMC unified storage platforms can use developing and bringing to market integrated restoration and recovery operations. to provide tangible benefits—helping free the solutions for Oracle Using SnapSure checkpoints, EMC Replication server processor, memory, and I/O channels environments. Manager enables quick replication and seeding from utility operations such as backup and Jeff Browning has been of production environments without reducing recovery, disaster recovery staging, testing and a technical evangelist for database server performance, helping to reduce development, and cloning. Oracle solutions for over per-processor licensing costs. Array-based Using Oracle DNFS with Dell/EMC platforms 20 years. He leads the EMC Proven Solutions synchronization of production environments can be an important step in developing an testing with Oracle and helps avoid the database log shipping overhead efficient architecture. DNFS is designed to VMware, and blogs at on the server processor. This array-based simplify network setup and management oraclestorageguy synchronization also helps protect the database by minimizing the need for administrative .typepad.com. from a variety of additional logical storage errors, tasks such as configuring network subnets, and production environments can quickly be performing LACP bonding, and tuning NFS restored to a specific point in time. parameters in Linux® operating systems. Load Learn more Organizations can use SnapSure writable balancing and high availability are managed Dell and Oracle: checkpoints to replicate one or many copies of internally within the DNFS client, an approach dell.com/oracle the production database. Changes can be rolled that can offer measurable performance gains. back into the production database if necessary. The use of a multi-protocol storage system, These replicas are well suited for use in quality like Dell/EMC unified storage, offers flexible Dell/EMC storage: dell.com/emc assurance, testing, and multiple reporting and scalable single-system configurations of environments, and for applying patches. mixed protocols and drives, including NFS, SnapSure creates a logical point-in-time image Common Internet File System (CIFS), Fibre Dell storage: dell.com/storage of a production file system (PFS) that reflects Channel, Internet SCSI (iSCSI), and Multi- the state of the file system at the time when the Path File System (MPFS) protocols along with checkpoint was created. SnapSure can maintain enterprise flash drives, Fibre Channel drives, and up to 96 PFS checkpoints. low-power Serial ATA (SATA) drives. It enables The basic principle of the SnapSure process an increasingly resilient and cost-effective is copy on first write. When a block is modified, architecture for nondisruptive migrations from a a copy containing the block’s original content SAN to NAS to support either cost-effective test is saved to a separate volume. Subsequent and development or mission-critical database changes made to the same block are not operations in growing organizations.

Taking an intelligent data management approach to archiving data

By Kay Benaroch and Joe Colucci

rowing data volumes are making data content (such as e-mail, rich media objects, management increasingly costly and Microsoft® Office PowerPoint® presentation files, labor intensive for IT departments. One and Web content) grows, and the popularity of Gprimary culprit is e-mail. As employees platforms such as Microsoft Office SharePoint® Server become increasingly mobile and e-mail becomes collaboration and document management software the preferred method of enterprise communication, increases, it can be difficult to determine the value e-mail volume is expanding exponentially. And of each piece of stored information—so enterprises organizations are keeping their e-mails around for often simply default to saving everything. a long time. According to the Forrester Q1 2010 However, for many organizations, only a small Global Message Archiving Online Survey, for example, portion of the data on primary storage is active at any 57 percent of survey respondents with currently given time. A lack of uniformity in applying existing deployed message archiving solutions retained e-mail data management policies can further compound for an average period of 5–10 years.1 the data growth problem. The result: all too often, a Collaborative work styles are also contributing company’s primary storage is overflowing with aged, to rapid data growth. As the volume of unstructured duplicate, or nearly worthless data.

1 “Regulatory and eDiscovery demands drive a growing message archiving market,” by Brian W. Hill with Matthew Brown and Sara Burnes, Forrester Research, Inc., March 17, 2010.

32 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Disk • Hierarchical storage Large disk management (HSM) • Content-addressable Fast or local tape quick data recall storage (CAS) • HSM to tape • Local storage • E-mail (auto-archiving) Remote tape or • Fixed media optical media • Remote copy • Files, documents, and images • Used where disk is • Long-term storage impractical • Local storage • Remote library • Files, documents, • Off-site tape storage • Remote copies and images • Local tape library • Remote tape library Archiving for space reclamation

Archiving for compliance Figure 1. Two categories for archiving data

At the same time, retention and deletion Archiving as a foundation requirements for compliance are becoming of intelligent data management increasingly complex. The retention period for a To address the challenges created by data specific file type depends on a combination of growth, enterprises must take a comprehensive governmental, industry, application, and business approach to IDM that spans the entire unit requirements. Organizations can face information life cycle, from creation and stiff penalties if they do not retain the proper distribution through use, maintenance, and information for the required period, and they can disposal. Key technologies for IDM include the also face liability if they retain certain data past its following: expiration date or are unable to find critical files or e-mail correspondence during litigation. • Object storage: Objects are files or file Archiving—moving valued data from high- collections that are enhanced by metadata cost transactional storage media to lower-cost, that includes an object ID. Enterprises can fixed, or read-only storage—helps enterprises leverage this standard to store, access, address these challenges. However, confusion and distribute digital content by creating about the differences between archiving and efficiencies, ease of access, and simplified backup processes can also create inefficiencies management throughout the life cycle of and unnecessarily increase costs. Archiving unstructured data (see the “Dell DX Object moves aging or infrequently accessed data off Storage Platform” sidebar in this article). of primary storage media to less-expensive • Deduplication: Deduplication technology storage media, thereby removing it from the plays a key role in data optimization by helping backup stream; backup simply creates a copy eliminate multiple copies of data at file, block, of the data on the secondary storage, without or sub-block levels to reduce the total capacity moving the original information. As a key required to store and protect data. part of a comprehensive intelligent data • Unified storage: Unified storage includes management (IDM) strategy, effective archiving data tiering and optimization in a single can help organizations increase efficiency array that provides an integrated platform for IT staff while cost-effectively meeting designed to simplify data management, and compliance requirements and streamlining offers integrated tiering and array-based backup processes. deduplication across multiple types of data.

Data archiving is a key pillar of interchangeably. However, backups IDM. By relocating little-used data from cannot act as archives, or vice versa. Dell DX Object primary storage to a long-term data Understanding the differences between repository on a separate physical tier of these two processes is critical to Storage Platform storage media, archiving helps free up optimizing efficiency and reducing the space on high-performance disks costs associated with IDM. During a The Dell™ DX Object Storage Platform for transactional data. Archiving falls into backup, the primary copy of the data supports archiving with policy-based two distinct categories (see Figure 1): remains in its original location after it management capabilities, including has been copied; each time a backup automated, object-level retention and • Data archiving for space reclamation: is performed, a new copy of the data is deletion based on metadata—helping This category requires policies and produced and stored. avoid the human error, access, and software to mark files for movement Archiving, in contrast, involves preservation problems that can increase from high-cost primary storage to moving the primary copy of the data risk and drive up costs. lower-cost (and lower-performance) off the primary storage media to a The platform is designed to intelligently storage media while still retaining more cost-effective storage platform. access, store, protect, and distribute fixed access to help meet enterprise Deduplication technologies can digital content. Automated management, requirements. Space reclamation complement both backup and archiving Serial ATA (SATA)–based storage, and a includes archiving of data such as processes by helping eliminate multiple modular scaling approach offer low total files, e-mail, or Microsoft SharePoint redundant copies of data. The archive cost of ownership. Self-healing functionality data stores that contribute to the rapid process can help alleviate backup issues continuously self-manages the integrity of growth of primary storage. by reducing the volume of data that stored objects to detect and repair errors, and • Data archiving for legal retention and must be backed up regularly. automatically reconfigures and regenerates regulatory compliance: This category Reducing the amount of primary objects without the need for IT intervention. requires legally approved policies that storage being backed up helps The DX Object Storage Platform enables direct data movement from primary ensure that backups are completed enterprises to keep more data online storage to a specified storage device within the available window. The and accessible than traditional storage for long-term managed retention. differences between archiving and platforms allow, because it doesn’t need This task can be handled with content backup become extremely important the conventional file hierarchies associated archiving or object-based storage and when an organization must perform with traditional storage area networks and automated policies. High-performance e-discovery in response to litigation. If network attached storage. An extremely large disk storage is not required, because the organization is asked to produce number of available addresses means that a data recall only occurs in response to data that IT leaders thought they had common pool of storage can support literally planned events or legal e-discovery, archived, and it turns out they had billions of objects. It provides continuous which allows for retrieval from offline actually backed up the data to tape, migration and investment protection with media that does not need to be IT personnel must begin going back nondisruptive security and capacity upgrades executed as rapidly. The storage device through their tapes to find the required as well as node replacement and retirement. must be read-only and must have the information—a process that may take It also supports selectable data immutability capability of enforcing retention limits months. Inefficient manual e-discovery for compliance needs. for files. It must be able to recognize, processes can, in turn, contribute to A large and ever-growing number of based on policy, when a file is no increased labor and litigation costs. independent software vendors support the longer required to be held, and then DX Object Storage Platform. This support delete or destroy the file. Expanding data management helps maximize an enterprise’s investment by strategies utilizing a common object-storage pool for Distinguishing archiving and Dell can help enterprises expand their multiple applications such as file, e-mail, and backup processes existing data management strategies medical record archiving software. In many organizations, backup and and strengthen their archiving practices archiving processes are discussed by applying automated policies to help

34 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. determine the relative value (relevance) of their the challenges. A Dell team performs an data. Object-based storage platforms used with assessment to capture the current state and applications from independent software vendor requirements, then designs an archiving services Comprehensive partners can help to identify and tag data objects catalog to guide the solution configuration and archiving for by scanning metadata during the process of implement Dell best practices for archiving. Microsoft Exchange creating data archives. This approach helps As this solution brief shows, IT departments avoid archiving unnecessary Taking a broad approach to intelligent the combination of the Dell data, simplifies data-store searches, and data management DX Object Storage Platform and NearPoint software from allows information to be retrieved, stored, and Although many organizations consider archiving Iron Mountain can provide an managed based on its attributes—that is, on the and backup processes to be interchangeable, efficient and effective way to store, manage, and protect information contained in metadata associated they are not. Both are key cornerstones of an user-generated, unstructured with the stored objects. IDM strategy. Through a variety of key services, Microsoft Exchange e-mail Dell also helps organizations reduce Dell can help enterprises expand their data content. complexity through policy-based automation that management strategies and strengthen archiving dell.com/downloads/global/ moves data into archive repositories designed practices to create an effective, efficient IDM products/pvaul/en/dell-dx -ironmountain-solutionbrief.pdf to reduce primary storage data stores. Dell approach to data management. unified storage solutions can support enhanced enterprise efficiency, while Dell’s incorporation Kay Benaroch is a senior marketing of data deduplication technologies with software consultant at Dell focused on data protection, and hardware helps to reduce the backup storage deduplication, and object-oriented storage, footprint, power costs, and wide area network and has more than 30 years of experience in marketing technology products. replication bandwidth requirements. Dell Data Management Consulting Services Joe Colucci is a global solution architect provide analysis to help determine when to on the Dell Global Infrastructure Consulting Services Data Management team. He has implement archiving and other IDM technologies; 15 years of IT experience, including 10 years Dell consulting experts can also help design and in storage consulting. implement archiving solutions. Data Management Assessment Services help organizations Learn more define their challenges and requirements through a workshop designed to highlight data Intelligent data management: management issues that could be compounding dell.com/datamanagement

Learn how new DC-powered DellTM PowerEdgeTM servers can address the need for more energy-efficient ways to run enterprise-class applications and data center operations. Free whitepaper: Enabling Greenfield Application Deployments with PowerEdge Technologies

For more information or to download the whitepaper: visit www.nei.com/dell or call +1.972.633.3400 Storage optimization

irtual machine (VM) sprawl can lead to Accelerate operations resource contention, duplicate data, inefficient utilization of processor cycles, with next-generation Vand inconsistent data protection. And because storage and virtualized environments are typically managed separately through independent storage efficiencies tools and consoles, it can be difficult for IT organizations to efficiently provision and configure By Annette Cormier, Eric Cannell, and Brad Bunce resources, troubleshoot performance bottlenecks, and perform capacity planning and trend analysis. Dell/EMC NS unified storage offers tight integration To help address these challenges within VMware virtual environments, Dell/EMC NS unified storage with VMware® vSphere™ performance, utilization, and offers tight integration with VMware vCenter™ recoverability features designed to increase quality of software with key features—including the EMC® Unified Plug-in for VMware; EMC Unisphere™ storage service and efficiency within virtualized environments. management software; optimization capabilities such as EMC Fully Automated Storage Tiering (FAST), FAST cache, and data deduplication and compression; and data protection with the EMC Celerra SnapSure™ and EMC Celerra Replicator™ applications. Through these and other key technologies, Dell/EMC NS unified storage can help simplify management, enhance efficiency and performance, streamline data protection and recovery, and accelerate VM deployments and backups.

Flexible storage management Because virtual infrastructures are dynamic, storage requirements for VMware virtualization vary across deployments, and can rapidly change as new VMs are spawned and as virtualization deployments move from test and development to mission- critical operations. As a result, organizations often need to support multiple file- and block-level storage protocols. Internet SCSI (iSCSI) may be Versatile storage desirable for leveraging staff with existing IP skill sets, while network attached storage (NAS)—using Multi-protocol Dell/EMC NS unified storage is Network File System (NFS) and Common Internet designed to maximize flexibility in virtualized File System (CIFS) protocols—can be effective for environments. organizations deploying virtual desktop infrastructure environments. Mission-critical operations, such as • Simplified self-service management and the online transaction processing used in trading and provisioning from the VMware console e-commerce, may have low-latency quality-of-service • Optimized performance with mission- requirements that would be best suited to the critical service-level agreements Fibre Channel protocol. • Network-efficient data protection to help The support for both file- and block-level meet rigorous requirements protocols in Dell/EMC NS storage enables

36 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Block over IP deployment VMware over ﬁle deployment Block over Fibre Channel

VM VM VM … VM VM VM VM VM VM VM VM VM VM VM VM … VM VM VM VM VM VM VM VM Application servers, Microsoft SharePoint® software, Microsoft® Active Directory® directory decision support systems, service, Microsoft Exchange Virtual desktops data warehouses

iSCSI SAN NFS or CIFS Fibre Channel Shared SAN templates

NAS

Fibre Figure 1. Dell/EMC NS unified storage provides a robust iSCSI Channel platform for VMware virtual environments

organizations to support a wide range of The plug-in also offers end-to-end mapping virtualization use cases within a single storage and visualization of VMs and the storage infrastructure (see Figure 1). Organizations can resources that support them, and supports both Simplifying use the protocols that they have experience block and file storage provisioning through NFS, storage with or those that are most appropriate for their CIFS, iSCSI, and Fibre Channel protocols. VMware management application needs, and can easily add or modify administrators can view and self-service provision storage based on usage and growth. the storage infrastructure from the vCenter Visit Dell online to learn more about Dell/EMC The EMC Unified Plug-in for VMware— console, including creating and deleting block- unified storage, view an available at no additional cost for Dell/EMC level Virtual Machine File System (VMFS) data EMC Unisphere demo, and register for local seminars NS unified storage and Dell/EMC CX4 Series stores and file-level NFS data stores. New data and on-demand Webcasts. storage—enables VMware administrators to stores are automatically created and added to the dellenterprise.com/ provision storage from the VMware vCenter VMware vSphere platform cluster. Administrators goto/emc console (see Figure 2). This plug-in provides can easily assign allocations at the cluster, ESX end-to-end mapping of VMs to the physical host, VM, or even VMware Distributed Resource storage resources that support them, enabling Scheduler (DRS) resource pool level, and can both storage and VMware administrators to view identify the connection between data stores, and manage the virtualized infrastructure through LUNs, targets, VMs, and storage. a single console. It also enables administrators Other key features include the following: to authenticate and connect to all Dell/EMC NS unified storage in the environment regardless of • Compression: Dell/EMC NS unified storage protocol; shows the correlation between logical helps reduce storage sprawl by offering units (LUNs), data stores, hosts, targets, and virtual both file- and block-level data compression. storage; and provides advanced details on how Compression can be activated within storage is being utilized within the environment. the VMware vCenter console, enabling

Targeted performance optimization VMware virtual environments often have a range of storage performance requirements within a single infrastructure. Dell/EMC NS unified storage offers several performance optimization features that enable VMware environments to cost-effectively and automatically meet varied performance requirements. EMC Unisphere extends integration with VMware environments through VMware vStorage APIs for Array Integration (VAAI), in which storage- related functions are offloaded from the VMware server to block-based Dell/EMC storage, including Dell/EMC NS unified storage with the Fibre Channel option and Dell/EMC CX4 Series storage. This approach enables increasingly efficient use Figure 2. The EMC Unified Plug-in for VMware offers a single-console view of end-to-end virtualized infrastructures of server resources to enhance performance and consolidation, hardware-accelerated copy administrators to compress and uncompress operations, and bandwidth and processor VMs in an NFS file store. Network-efficient utilization. backup and recovery is achieved with the Dell/EMC NS unified storage supports a combination of compression and deduplication, wide range of drive types to facilitate targeted, which helps considerably reduce the amount of cost-effective performance, including enterprise data transferred over the network. flash drives (EFDs), also known as solid-state • Deduplication: Duplicate files (or redundant drives; Serial ATA (SATA) drives; and Fibre Channel data) and infrequently used files can consume drives. EFDs are designed to offer extremely extra capacity and network bandwidth, both high-performance and low-latency storage Deploying during storage operations and during backup that is well suited to the most performance- EFDs in Oracle and recovery operations over a wide area sensitive applications. High-capacity SATA II environments network (WAN) or local area network (LAN). drives offer cost-effective, high-capacity storage for In this on-demand Dell/EMC NS unified storage offers file-level applications such as archiving software that are not presentation, Eric Cannell deduplication, which, when combined with as performance intensive as applications that require discusses how EFDs can help organizations create compression, can help reduce file data capacity high I/O workloads. Fibre Channel drives can offer a high-performance, requirements by up to 50 percent. reliable, high-performance storage for applications efficient, scalable storage infrastructure to support • Automated volume management: The EMC that require extremely fast response times. Oracle® environments. Unified Plug-in for VMware helps automate the To help optimize performance in dynamic

dellenterprise.com/goto/ process of creating NFS data stores by enabling VMware environments, some Dell/EMC storage emcoracle administrators to easily provision a file system, arrays also offer EMC FAST technology. FAST can optimized by workload, in just a few simple steps. automate data movement across drive types within an array based on preset policies and I/O usage To offer storage administrators comprehensive patterns. Heavily accessed data or data with very visibility into virtual infrastructures, the EMC high performance requirements can be moved to Unisphere storage management tool is tightly high-performance storage, such as EFDs, while integrated with VMware platforms. Storage infrequently accessed data can be moved to more administrators can use it to see the end-to-end cost-effective storage, such as SATA drives. mapping of VMs to storage resources as well as FAST supports automated movement of data additional information such as disk consumption at the sub-LUN level as well. This optimization and capacity utilization for individual VMs. feature is especially beneficial in VMware

38 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. environments, because many VMs and data technology helps minimize disk space and stores may reside on a single LUN but have very enhance performance. Full cloning enables the different performance needs. Using this feature creation of a full copy of the VM in the same data can help organizations meet high performance store or in an alternate data store. Rapid cloning requirements cost-effectively and efficiently. of VMs is important for simplifying virtual desktop VMware environments are also prone to deployments and when scaling virtualized unexpected spikes in application workloads that infrastructures from test and development to can overwhelm cache capacities. To help maintain mission-critical virtualization operations. performance and ensure that response times EMC Celerra Replicator software supports and service-level agreements are maintained automated, asynchronous, multisite replication. during these spikes, Dell/EMC NS unified storage Celerra Replicator creates point-in-time, read- includes FAST cache as a flash-based extended only, and read/write copies of production file Annette Cormier is cache. FAST cache allows flash drives to be used systems or LUNs on local or remote storage a solutions marketing manager for Dell/EMC as an extension of the system cache, and is fully systems. It supports up to 1,024 replication storage solutions with read/write capable. This automatic absorption of sessions, sending only changed blocks over 20 years of experience in excess workload helps to minimize performance the network, and does not affect VMware developing and bringing degradation because of unexpected spikes in virtualization performance. A recovery point to market enterprise storage, network workload, while helping avoid the need to expand objective is set for each file system to be management, and expensive memory cache. replicated. Celerra Replicator then manages the security products. transmission of changed data to the target file Eric Cannell is a product Streamlined data protection and recovery system. Celerra Replicator is designed for ease marketing manager for Enterprise VMware virtualization deployments of use, scalability, and flexibility to help meet the Dell/EMC storage systems generally have rigorous data protection and most demanding disaster recovery infrastructure with many years of experience in developing recoverability requirements. To help ensure data requirements. enterprise technology protection and recoverability, Dell/EMC NS unified Automated disaster recovery is also supported products. storage provides robust, built-in data protection through the EMC VMware vCenter Site Recovery features including automated snapshot, Manager (SRM) Failback plug-in, which leverages Brad Bunce is the replication, and recovery capabilities. Celerra Replicator to enable end-to-end disaster technical marketing director for EMC EMC Celerra SnapSure software, included with recovery of VMware infrastructures from the midrange storage Dell/EMC NS unified storage, can create application- VMware vCenter SRM application. It also provides solutions. He has consistent point-in-time snapshots, including data replication of all or part of the virtualized 17 years of IT experience writable snapshots for test and development. infrastructure, supports both Dell/EMC NS in technical presales, IT management, and SnapSure supports built-in automated recovery. and Dell/EMC CX4 storage systems, enables marketing. Administrators can create up to 96 read-only file automation of the recovery process, and supports system copies, which are immediately available to disaster recovery simulation and testing. end users and applications; administrators can also Learn more have 16 copies available for read/write. End users Efficient unified storage Dell/EMC storage: can access the copies to perform self-service file- for virtualized infrastructures dell.com/emc level recovery through copy-and-paste operations. The ability to support self-service storage The copies are also available to IT staff to perform management and provisioning—and maintain Is Dell/EMC right complete file system recoveries. Because the copies data protection and recoverability in enterprise- for you?: store only changed data, they are generally much level virtualized environments—is available dellenterprise.com/ smaller than the production file systems; only now. Through tight integration with the goto/emc-jellyvision current data is kept in the production file system. VMware vSphere platform, Dell/EMC NS unified SnapSure also provides fast VM cloning storage helps dramatically streamline storage functionality, which enables administrators to management and data protection within VMware create pointer-based clones that are the size infrastructures while enhancing quality of service of unique data only. This pointer-based copy for mission-critical operations.

Best practices for deploying iSCSI SANs in DSS database solutions

By Suresh Jasrasaria, Ananda Sankaran, and Chris Almond

Following key best practices can help organizations create scalable, high-performance decision support systems or data warehousing solutions based on the Microsoft® SQL Server® 2008 database platform and Dell™ EqualLogic™ PS Series storage arrays.

ffective decision support systems • Maximize end-to-end solution performance to (DSSs) or data warehousing (DW) confirm best practices for system configuration database solutions require scalable in each component layer E storage platforms that offer high levels of performance and capacity. DSS applications As the results demonstrate, following best are typically designed to support long-running, practices and taking advantage of flexible complex analytical query activities using very large EqualLogic storage can help organizations design data sets. The data I/O pattern from storage to and deploy scalable, high-performance DSS or server is predominantly sequential and usually DW solutions in their own environments.1 consists of large blocks—typically 512 KB or 1 MB in size. When designing such a system, IT DSS test environment administrators must consider the availability, The test environment included two scalability, and performance of each component configurations, both based on Dell EqualLogic layer within the solution stack, as well as how PS6010XV arrays, each of which had sixteen 450 these characteristics affect the overall solution. GB, 15,000 rpm Serial Attached SCSI (SAS) drives This article describes best practices for in a RAID-50 configuration. The first configuration deploying DSS or DW solutions based on used a single EqualLogic PS6010XV array for SQL Microsoft SQL Server 2008 using Dell EqualLogic Server data and another array for SQL Server PS Series Internet SCSI (iSCSI) storage area transaction logs; the second configuration used network (SAN) arrays. To demonstrate these three arrays for SQL Server data and one array practices, in January 2010 Dell engineers for SQL Server transaction logs (see Figure 1). deployed and tested two DSS configurations with The arrays were connected to a Dell the following design goals: PowerEdge M710 blade server with two quad- core Intel® Xeon® X5570 processors at 2.93 GHz; • Test a heavy DSS workload with high I/O 32 GB of RAM; two 73 GB, 15,000 rpm SAS drives; requirements against Dell EqualLogic the Microsoft Windows Server® 2008 Enterprise PS Series 10 Gigabit Ethernet (10GbE) OS; and the 64-bit version of Microsoft SQL arrays, Dell PowerEdge™ servers, and Dell Server 2008 Enterprise with Service Pack 1. The PowerConnect™ 10GbE switches test team had selected the PowerEdge M710

1 For the full version of this guide, see “Deploying Microsoft SQL Server 2008 based decision support systems using Dell EqualLogic 10GbE iSCSI storage,” by Ananda Sankaran, Dell Product Group, March 2010, available at equallogic.com/resourcecenter/assetview.aspx?id=9259.

40 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Dell PowerEdge M710 blade server Fabric A LAN on Fabric B Fabric C Fabric A Fabric B Fabric C Motherboard (LOM) mezzanine mezzanine LOM mezzanine mezzanine Broadcom Broadcom Broadcom Broadcom Broadcom Broadcom NetXtreme II 5709 NetXtreme II 57110 NetXtreme II 57110 NetXtreme II 5709 NetXtreme II 57110 NetXtreme II 57110

GbE GbE 10GbE 10GbE 10GbE 10GbE GbE GbE 10GbE 10GbE 10GbE 10GbE

Dell PowerEdge M1000e midplane

Dell 10GbE 10GbE Dell 10GbE 10GbE PowerConnect pass-through pass-through PowerConnect pass-through pass-through M6220 module module M6220 module module

Server LAN Dell PowerConnect Dell PowerConnect 8024F 8024F

Controller 0 Controller 1 Controller 0 Controller 1 Controller 0 Controller 1 Controller 0 Controller 1

10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE 10GbE

Dell EqualLogic PS6010XV Dell EqualLogic PS6010XV Dell EqualLogic PS6010XV Dell EqualLogic PS6010XV

SQL Server data SQL Server logs

Figure 1. Second configuration used in the DSS test environment as the database server in order to take Best practices for DSS component throughput among the supported hard disk advantage of its expanded I/O capability. configuration drive types. In environments requiring lower To create the 10GbE iSCSI storage The test environment was designed to levels of I/O throughput, administrators network, the test team installed two follow best practices when configuring could consider using 10,000 rpm SAS drives Broadcom NetXtreme II 57710 10GbE the components of the DSS solution stack. in an EqualLogic PS6010X array or Serial mezzanine network interface cards (NICs) This section outlines some of the key ATA (SATA) drives in an EqualLogic PS6010E on the server in fabric B, and connected recommendations that IT administrators can array. The appropriate drive size depends on these NICs to external PowerConnect take into account when designing a scalable the capacity requirements of the database. 8024F switches through the pass-through solution in their own environments. Select an appropriate RAID policy. modules installed on the PowerEdge The test environment used RAID-50 on M1000e modular blade enclosure. (A Dell Dell EqualLogic storage array configuration each array across 14 disk drives, with the PowerConnect M8024 switch can also be The following are recommended best remaining 2 disk drives set as hot spares. used as an alternative to the pass-through practices for EqualLogic storage array It is a best practice to use RAID-50 modules.) A PowerEdge M610 blade configuration as part of a DSS solution. because DSS environments are highly server ran the DSS query simulation tool Select an appropriate disk drive type read intensive and require large storage (Quest Benchmark Factory) and hosted and size. The test environment used capacities. RAID-10 is also recommended the EqualLogic SAN HeadQuarters (SAN EqualLogic PS6010XV storage arrays for environments that include a mix HQ) monitoring application. A modified configured with sixteen 450 GB, 15,000 of both read and write operations. benchmark similar to TPC-H was used to rpm SAS drives, because 15,000 rpm SAS Select an appropriate storage pool model the DSS workload. drives offer the highest level of disk I/O configuration. The test team created

two storage pools in each test configuration: increase SAN throughput by allowing additional one pool hosted the SQL Server data volumes, data to be transferred in each Ethernet frame. including the tempdb volumes, and the other The jumbo frame size on the switch ports hosted the SQL Server transaction logs. The must match or exceed the settings at the end- transaction logs were physically separated from device ports (the server and storage NICs). the DSS database disk drives to provide high Generally, end devices conform to a jumbo levels of data protection and recoverability. frame setting of at least 9,000 bytes and switches DSS workloads are primarily read intensive and of up to 9,216 bytes. It is a best practice to typically generate minimal transaction log activity. enable flow control (both TX and RX) and jumbo Thus, in smaller DSS database environments, frames on all switch ports, including virtual port the database and transaction log volumes can channels created for link aggregation groups. be typically hosted on the same pool without Pay attention to Spanning Tree Protocol sacrificing performance. SQL Server mirroring (STP). The test team disabled STP on all switch or replication solutions can also be considered ports connecting to end devices (the server and for additional recoverability in such scenarios. storage controller NICs), and enabled the PortFast setting on those switch ports. Because NIC and SAN configuration storage controller ports are endpoints, they The following are recommended best practices cannot create loops. Enabling the PortFast setting for SAN configuration as part of a DSS solution. avoids the STP forwarding-state delay time that Use redundant switch paths between servers occurs when a port becomes active, which helps and storage arrays. The test configurations used accelerate initialization of new iSCSI sessions redundant switch paths in the SAN to help ensure when ports on an existing iSCSI connection path uninterrupted access to storage controllers in the change status. event of a switch or interconnect cable failure or Select appropriate physical interconnects. scheduled switch downtime for maintenance. PowerConnect 8024F switches support small The following best practices are recommended form-factor pluggable + (SFP+) ports for 10GbE when designing an iSCSI SAN for a DSS solution: connectivity. Support for 10GBase SFP+ is typically provided with 10GbE switches. Either SFP+ • Equally distribute the NIC ports installed on the transceivers with optical cables or SFP+ copper server for iSCSI SAN connectivity across the cables can be used for interconnecting switches redundant switches or fabric. and for connecting end devices to the switches. • Equally distribute the Ethernet ports from Connection distance and cost determine the each controller of the storage array across the appropriate type of physical connection. redundant switches or fabric. • Interconnect the two switches with a sufficient Dell PowerEdge server configuration number of inter-switch links to support the The following are recommended best practices workload. for PowerEdge server configuration as part of a DSS solution. It is also a best practice to always isolate the Configure server hardware to meet SAN so that the iSCSI switching infrastructure is performance and capacity requirements. In dedicated to storage traffic. the test environment, the PowerEdge M710 Use flow control and jumbo frames. Flow database server was configured with two quad- control increases the ability of the SAN to core Intel Xeon X5570 processors and 32 GB of regulate packet flow between nodes (servers, RAM to meet the needs of the workload under switches, and storage) during conditions of test. The Dell SQL Server Advisor tool (available high I/O traffic. Jumbo frames are Ethernet at dell.com/sql) can provide a starting point frames that have a maximum transmission unit for sizing server configurations to support SQL larger than 1,518 bytes. Jumbo frames can help Server DSS workloads.

42 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Think beyond processor and memory. 3.5 In the test environment, each NIC was 3.0 connected to the two blade sockets 2.5 internally through an x8 PCI Express (PCIe) 2.0 interface. This eight-lane interconnect path 1.5 can provide up to 16 Gbps of PCIe 1.1– compliant bandwidth per mezzanine NIC. 1.0 (The PowerEdge R710 supports PCIe 2.0; 0.5 (normalized linear scale units) using PCIe 2.0–compliant NICs would Instantaneous peak throughput 0 Conﬁguration 1 Conﬁguration 2 increase this bandwidth further.) It is a best (one array for SQL Server data) (three arrays for SQL Server data) practice to choose servers with appropriate PCIe technology, a sufficient number of Figure 2. Instantaneous NIC peak throughput for each test configuration PCIe slots, and a sufficient number of PCIe lanes to provide the data transfer 70 bandwidth required by the DSS workload. 60 Properly size the SAN connection 50 paths. It is a best practice to provision the 40 quantity (at least two for redundancy) and 30

connection speed of the SAN NIC ports on (percent) the server based on workload throughput 20

requirements, and to provision separate utilization processor erage 10

NICs for the server LAN connections. Av 0 Conﬁguration 1 Conﬁguration 2 (one array for SQL Server data) (three arrays for SQL Server data) OS configuration The following are recommended best Figure 3. Average server processor utilization for each test configuration practices for OS configuration as part of a DSS solution. Align disk partitions with storage stripe Utilize NIC offload capabilities. The Use the EqualLogic MPIO DSM to boundaries. If disk partitions do not align test team used Broadcom NetXtreme II optimize the MPIO configuration. The with external storage RAID striping scheme 57710 NICs in TCP/IP Offload Engine EqualLogic MPIO DSM is provided in the (that is, the block size), then the system (TOE) mode. Enabling this mode on the EqualLogic Host Integration Tool Kit for generates more I/O operations than in an NICs to work in conjunction with the Microsoft Windows® operating systems, optimal configuration. By default, Windows Microsoft iSCSI Software Initiator, Microsoft and can provide enhanced performance Server 2008 aligns partitions with an offset Multipath I/O (MPIO), and the EqualLogic for bandwidth-intensive applications of 1,024 KB, and thus can accommodate MPIO Device Specific Module (DSM) helps such as SQL Server DSS workloads. all RAID stripe sizes up to 1,024 KB. For provide optimal performance for SQL The test environment used the default previous-generation operating systems, Server DSS workloads. As an alternative to values for the number of connections the best practice is to manually align the TOE, the network card can also be set up per member (two) and the total number data partitions with an offset suitable in the iSCSI Offload Engine (iSOE) mode. of connections per volume (six). to the RAID stripe size using a utility such Performance of offload engine modes as DiskPart. from different NIC vendors may vary. Microsoft SQL Server 2008 configuration Use flow control and jumbo frames. Disable NIC teaming and unused The following are recommended As with the SAN configuration, it is a NICs. It is a best practice to disable best practices for SQL Server 2008 best practice to enable flow control (TX NIC teaming on all ports used for iSCSI configuration as part of a DSS solution. and RX) and jumbo frames on the server connections when MPIO is enabled. Properly size the storage volumes NIC ports connecting to the SAN, and Disabling unused NIC ports can help for each database I/O component. SQL to match these settings with the switch simplify systems management and reduce Server data layout on the storage volumes and array controller port settings. configuration errors. is an important factor for optimal operation

300 partitioning scheme was based on a particular table column that contained the most common 250 field in the user queries. 200

150 DSS test results Figure 2 shows the relative instantaneous 100 peak throughput at the server NICs (receive) 50 in each configuration, as measured using the

Query time (minutes) response 0 Microsoft Windows Performance Monitor tool Conﬁguration 1 Conﬁguration 2 while the DSS queries were executing. The (one array for SQL Server data) (three arrays for SQL Server data) maximum throughput observed at the server Figure 4. Total query response time for each test configuration NICs scaled almost linearly from the single-array configuration to the three-array configuration. This scalability was due to the increased storage of DSS workloads. Storage volumes must be bandwidth available (three times the number of appropriately sized for capacity and performance storage array 10GbE controller ports) and the to host the database data, transaction logs, increase in available storage resources (three and tempdb databases, respectively. Because times the number of storage controllers and DSS workloads often create large temporary disk drives). data objects during query processing, it is Figure 3 shows the average processor critical to size the tempdb appropriately. utilization on the database server during each Match RAID levels to DSS I/O components. test; Figure 4 shows the query response times Different database types can generate different measured at the test client for completion of all I/O patterns in a DSS environment. The DSS query streams. These results illustrate that as Suresh Jasrasaria is a product recommended RAID levels for each volume type more I/O bandwidth is available (because of the marketing senior consultant in the Dell Enterprise Storage Product are as follows: increased storage resources), the database server Group, and has worked in the data was able to reduce query response times. More storage industry for more than • Database volumes: RAID-50 is recommended processor cycles were utilized to process the data 20 years. to optimize for both capacity and performance. arriving at a higher rate with three arrays. Ananda Sankaran is a senior • Tempdb volumes: If write activity is low, development engineer on the Dell RAID-50 is recommended to optimize for both Scalable, available, high-performance Storage Infrastructure and Solutions team, and has more than 9 years of capacity and performance. If write activity is design experience in software engineering, high, RAID-10 is recommended. Scalability and availability are key factors of an systems engineering, and solutions • Log volumes: In most cases, RAID-50 is effective DSS or DW solution. Adhering to the best architecture. recommended to optimize for both capacity practices outlined in this article for integrating Dell

Chris Almond is a senior consultant and performance. If the DSS database refreshes EqualLogic PS Series iSCSI SAN arrays into the responsible for development of include very large bulk updates, however, then design can help IT administrators create a scalable publications produced by the Dell RAID-10 is recommended. high-performance infrastructure that can grow as Storage Infrastructure and Solutions team. the needs of the organization change. Use SQL Server table partitioning to help increase query processing performance. Table

partitioning is a recommended best practice to Learn more enhance query processing, particularly when implemented on frequently accessed large tables Dell and Microsoft SQL Server: and when based on the most commonly queried dell.com/sql table field. In the test environment, the largest Dell EqualLogic PS Series: table was partitioned into four subsets, each dell.com/psseries residing on its own file group and volume; this dell.com/equallogic

irtualization is a widely adopted strategy Rightsizing storage for solving many of the challenges IT organizations currently face, such as V simplifying infrastructure, optimizing for small to data management, and using resources efficiently. Organizations are deploying virtualization across midsize virtualized server and storage platforms to help reduce capital and operational costs while increasing IT productivity. environments Research supports the increasing momentum of virtualization adoption: according to Gartner, “only By John Mannix and Kyle Walczak 16 percent of workloads are running in virtual machines today, but Gartner predicts that this will rise to around 50 percent of x86 architecture Virtualization is growing rapidly in small and midsize server workloads by the end of 2012, representing organizations. Dell™ PowerVault™ MD32x0 SAS and approximately 58 million deployed machines.” Tom Bittman, vice president and distinguished analyst at PowerVault MD32x0i iSCSI arrays are designed to Gartner, says, “While large enterprises were quick to meet these organizations’ performance, capacity, leverage virtual machines to reduce server sprawl and power costs, as well as conserve data center flexibility, and ease-of-use requirements. space, small business started late on virtualization. However, by year-end 2010, enterprises with 100–999 employees will have a higher penetration of virtual machines deployed than the Global 500. For years the entry point was simply too high for small enterprises, but increased competition by server vendors has enabled smaller firms to embrace virtualization.”1 The data storage needs of small and midsize organizations are growing, and virtualization is helping to improve the cost-efficiency of storage systems that address these needs. Storage virtualization is increasingly deployed as a key component in server virtualization projects, helping consolidate storage resources and reduce infrastructure investment and operating costs. Exceptional performance As small and midsize organizations increasingly deploy virtualization, the demands placed on The Dell PowerVault MD32x0 series and their storage systems that support the virtualized PowerVault MD32x0i series incorporate a range workloads also continue to increase. These of features to help meet the needs of small and organizations must support the growing deployment midsize organizations. of virtual machines using fewer resources than large enterprises. Strategies often include streamlining • Scalability to grow as capacity needs administration to offset limited IT staffing, and increase looking for ways to obtain additional leverage from • Flexible deployment options to fit in with virtualization technology investments. existing platforms

• Integrated functionality including data 1 “Gartner says 16 percent of workloads are running in virtual machines today,” press release, Gartner, Inc., October 21, 2009, gartner.com/it/ recovery and protection page.jsp?id=1211813.

With data volumes continuing to Other core storage requirements for expand rapidly, finding smaller-scale small and midsize organizations include Small but powerful storage arrays to support the following: virtualization becomes increasingly environments, important. Several considerations and • Performance: Organizations need big demands trade-offs are involved in deploying scalable application performance storage systems for virtualization projects. within budget. Performance should Small and midsize organizations often have be sufficient to meet the demands the same performance demands as large Defining the requirements of the large database applications enterprises, but may not have the same level of virtualized storage used by many midsize organizations. of expertise or the budget to deliver that The storage systems needed to support The ongoing proliferation of data performance effectively. Dell PowerVault growth in small and midsize organizations types—including media such as audio, MD32x0 series Serial Attached SCSI (SAS) are different from large enterprise video, and high-resolution images— arrays and PowerVault MD32x0i series systems. In addition to right-sized capacity further increases need for processing Internet SCSI (iSCSI) arrays can provide a and performance, small and midsize performance. variety of benefits for these organizations. organizations require storage that is cost- • Capacity: Small and midsize effective and based on open standards organizations often require enough The PowerVault MD32x0 series array with so that it can be deployed without capacity to store an estimated range four 6 Gbps SAS host ports per controller costly changes to existing environments. of 100–200 TB of data. Of course, offers the following advantages: Systems should also be highly flexible, the amount of data a fast-growing supporting a variety of virtualization organization stores today could be Highly available, dedicated hardware technologies, RAID levels, and drive a fraction of what it will need to platform to run a variety of applications types so that each organization can store a year from now. A modular Cost-effective, high-performance meet its specific requirements and fully architecture enables the organization virtualization for up to four servers utilize existing infrastructure. to scale efficiently and add capacity Systems that offer a choice of in cost-effective increments. Shared storage for clustered server Serial Attached SCSI (SAS) or Internet • Ease of use: Streamlined applications, designed for ease of use SCSI (iSCSI) host interface connectivity administration is essential for support multiple deployment options organizations that lack large IT staffs. The PowerVault MD32x0i series array and provide the flexibility that small Storage arrays must be easy to deploy with four iSCSI host ports per controller and midsize organizations need (see and manage, and provide built-in provides the following advantages: Figure 1). SAS connectivity is well intelligence to help avoid costly staff suited for deployments in which expansion or outside consultants. Cost-effective, high-performance bandwidth and price/performance Integration with leading hypervisors is iSCSI storage for growing virtualized are key considerations and when the also required to help reduce manual environments organization is directly connecting setup and make the most of the storage to multiple servers. Arrays virtualization investment. Consolidation to virtualize numerous based on iSCSI technology, on the infrastructure applications onto a other hand, are an excellent choice Assessing Dell PowerVault arrays reduced number of servers, helping when an organization is centralizing Dell PowerVault MD32x0 series SAS- improve utilization of IT resources storage for multiple servers on one based arrays and PowerVault MD32x0i Ethernet storage area network (SAN). series iSCSI-based arrays are specifically Increased application availability and Because iSCSI-based arrays work with designed to provide storage virtualization decreased backup windows cost-effective, well-understood Ethernet for small and midsize organizations technology, they can help reduce costs (see the “Small environments, big for additional training as well as for demands” sidebar in this article). These hardware such as switches and other arrays support storage virtualization infrastructure components. deployments with integrated functionality

46 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. VM VM VM VM VM VM 12 or 24 hard drives, respectively, to support a Host 1 Up to 32 hosts Host 32 maximum of 96 drives. PowerVault MD32x0 and iSCSI initiator iSCSI initiator PowerVault MD32x0i arrays support multiple network network virtualization technologies, including VMware® interfaces interfaces IF1 IF2 IF1 IF2 ESX and VMware vCenter™, Microsoft® Hyper-V™, and Citrix® XenServer™ software, enabling organizations to continue using existing platforms. Switched Virtualization and consolidation help simplify Gigabit Ethernet network storage management, and the PowerVault MD32x0 series and PowerVault MD32x0i series arrays can make management even simpler with P0 P1 P2 P3 P0 P1 P2 P3 PowerVault Modular Disk Storage Manager, a Controller 0 Controller 1 built-in application designed for easy interaction

Eth Eth with the system regardless of the user’s level of Dell PowerVault MD3200i array familiarity with storage infrastructure. The software (duplex mode) features an enterprise window that monitors multiple systems through a single interface as well as wizard-based array management to Management help streamline configuration. It can also detect IP network cloud problems and launch an automatic Recovery Guru Using Modular to help troubleshoot and resolve the problems Disk Storage using a built-in knowledge base. Manager Figure 1. Example Dell PowerVault MD3200i iSCSI array configuration in a virtualized environment Starting small with a scalable foundation PowerVault Modular Disk Storage Manager features Virtualization is a valuable IT strategy, and storage system status monitoring such as data recovery, online data migration, is an important part of that strategy for small and for multiple arrays, an image view of the system that application awareness, data protection, and failover. midsize organizations. Dell PowerVault MD32x0 provides status information For example, these arrays can take point-in-time series SAS arrays and PowerVault MD32x0i series for key components, and quick and intuitive steps for snapshots of data for backup and make full copies iSCSI arrays are designed for these environments— creating a new virtual disk of virtual disks for decision support or software providing a rightsized, automated platform that and mapping it to a host. For more information, including a development testing. Data from key applications allows organizations to start small, minimize video demonstration, visit the can be restored back to a specified point in time in risks, and establish a solid foundation on which “learn more” links below. the event of a failure. Additionally, self-encrypting to grow. drives (SEDs) are designed to automatically make data unreadable by unauthorized persons if a drive John Mannix is a marketing manager on the is removed from the array. Dell Storage team and was previously with Gemalto, Hyperformix, and Cirrus Logic. The arrays are designed to provide the performance and scalable capacity that small and Kyle Walczak is a product marketing manager midsize organizations require. The PowerVault at Dell and has been in the storage industry for MD32x0 series provides up to twice the 13 years. performance of previous-generation PowerVault storage arrays, and even greater improvement Learn more can be achieved with solid-state drives and the optional High-Performance Tier firmware Dell PowerVault MD32x0 series: upgrade. Administrators can easily add capacity as dell.com/md3200 needs grow by adding hot-pluggable PowerVault MD1200 and PowerVault MD1220 expansion Dell PowerVault MD32x0i series: enclosures, with each enclosure holding up to dell.com/md3200i

Robust data protection in a virtual world

By Ed Casmer

Working with traditional backup tools in virtualized environments can be time-consuming and difficult. Symantec™ data protection software was designed with virtualization in mind—providing intelligent backups, granular file recovery, and more.

irtualization can bring numerous Even today, backup tools not specifically designed benefits to the data center, including for virtualization can have multiple limitations server consolidation as well as when used in virtualized environments—requiring V enhanced utilization, availability, administrators to take VMs offline during backup; and disaster recovery. Consolidation alone can restore entire VMs simply to recover a single file; provide substantial cost savings, yet consolidating perform lengthy, separate backup processes many virtual machines (VMs) onto one physical when running software such as the Microsoft® Symantec data system increases not only utilization, but also Exchange, SQL Server®, and Active Directory® protection in risk: a single physical hardware failure can bring platforms inside VMs; and more. action down multiple VMs. And the greater the density, In some scenarios, such a model can still the greater the risk—a lost host server could inhibit make sense. But as hypervisors have evolved, Visit Symantec online to view detailed video productivity for multiple departments for as long virtualized environments have become scalable, demonstrations and to learn as it takes to recover the virtual environment and high-performance platforms that support an more about Symantec data protection for virtualized the individual VMs. increasing number of powerful applications. environments. Because of these considerations, Practices once frowned upon—such as running

dell.symantec.com/psq organizations today are looking for efficient Exchange, SQL Server, and Active Directory backup and recovery software that can provide in a VM—are now generally accepted. Add to productivity benefits and cost savings on par this the growing density needs of data centers with those provided by server virtualization. around the world, and the need for finely tuned, Symantec data protection software such as the virtualization-aware backup software becomes Backup Exec™ 2010 and Veritas™ NetBackup™ 7 clear. Backup administrators and IT professionals applications can help meet these needs, providing don’t want to protect the physical and virtual administrators with powerful, flexible tools for aspects of environments at the price of vendor or backing up and recovering critical VMs and core product sprawl, nor do they want to protect VMs enterprise applications. running core business applications differently than they do VMs running Web servers, print servers, Designing for virtualization and similar basic services. When virtualization was introduced to the data Symantec data protection software is designed center, data protection was about protecting to meet both the virtual and physical needs of the the VMs. First-generation backup applications entire enterprise in a single package by leveraging were not tuned for virtualization; they simply Symantec’s innovative Granular Recovery expected an agent to be included in each VM. Technology (GRT). GRT helps to save time and

48 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Deduplicated storage VM Symantec VM VM Backup Files Files Files Virtual tape VM VM VM Exec 2010 library (VTL) App App App or Veritas Disk NetBackup 7 OS OS OS Tape

Microsoft Hyper-V or VMware Comprehensive In-line Support for Ability to write to virtualized servers automated ﬁle rapid search and restore from backup discovery and recovery of any supported individual ﬁles destination

Figure 1. Symantec data protection software helps simplify and automate backup and recovery in virtualized environments

reduce costs by allowing administrators to restore and Active Directory user accounts without individual files and folders within a VM from a requiring an application-agent-specific backup. single-pass image backup of the entire VM (see • Newly added or created VMs can be Figure 1). The VM need not be taken offline, nor is automatically included in backups. a separate file-by-file backup on the VM needed to restore an entire VM or individual files from Deploying comprehensive data protection within a virtual disk (.vmdk or .vhd). Additionally, As virtualization technologies have matured in Symantec software enables granular recovery scope from test and development applications to from an original image-based backup of a VM include today’s complex production environments, running software such as Exchange, SQL Server, Symantec data protection software has remained or Active Directory, without requiring separate at the forefront of the virtualization adoption curve. databases or application-agent-specific backups. By deploying tools such as Backup Exec 2010 and Symantec data protection software also offers Veritas NetBackup 7 in virtualized environments, several other key benefits: IT administrators can create a versatile, high- performance approach to protecting VMs and • Backups for Microsoft Hyper-V™ and VMware® core applications as part of a comprehensive virtualized systems and physical systems are enterprise backup and recovery strategy. provided in a single package. • Block-level as well as native file-level Ed Casmer is a technical strategist in the incremental and differential backups can Partner Alliance Group at Symantec dedicated to the Dell Alliance. help reduce backup time while also providing flexible backup policies. Learn more • Support for source and target deduplication

can dramatically reduce the amount of backup Symantec Backup Exec: data and help greatly increase storage density. backupexec.com • Single-pass backup provides both full and object-level restore, allowing Exchange, Veritas NetBackup: SQL Server, and Active Directory software symantec.com/netbackup running inside a VM to be protected while online and running. Administrators can Dell and Symantec: recover individual objects such as Exchange dell.com/symantec e-mails and mailboxes, SQL Server databases, symantec.com/dell

n the new virtual era, organizations need to get a lot more for a lot less. Technologies such as virtualization have provided some improvements Ibut have also added to the complexity of the environment. Increasingly, organizations must manage two infrastructures—physical and virtual— and have created “virtual machine sprawl.” In effect, one type of management complexity has been substituted for another. The Dell Virtual Integrated System (VIS) helps meet these challenges. VIS addresses IT efficiency in the virtual era by effectively managing technology as well as the people and processes that manage the technology. Efficiency is no longer confined to just hardware, and moving beyond hardware efficiencies is what defines the virtual era. VIS is designed to facilitate the data center transformation from statically to dynamically managed, and to address the operational challenges Building the created by this transformation. It streamlines key activities in the data center using an open dynamic data center architecture that seamlessly integrates into the existing environment. It extends the benefits By M. Consuelo Ortiz of virtualization while increasing infrastructure efficiency by helping reduce the time wasted on repetitive tasks. These advantages translate As a key component of the Dell Efficient Data into rapid, cost-effective resource provisioning and workload deployment, as well as increased ™ Center strategy, the Dell Virtual Integrated System efficiencies in IT processes, cycle times, and service architecture helps organizations move from a static levels—without compromising choice or simplicity. to a dynamic data center and increase efficiency in Dell VIS architecture the virtual era by effectively managing technology Data centers are heterogeneous, typically populated and the people and processes that manage the by a variety of server models, network switches, and storage systems from a range of vendors. technology. The adoption of server virtualization has added another heterogeneous component to the mix: hypervisors. These infrastructure components come with their own management tools, but the tools are typically not well integrated—which can lead to organizational silos and a plethora of management systems that data center managers must learn and support. This lack of integration can also result in diminished efficiency and increased response times. The open Dell VIS architecture helps address these problems by providing integration, automation, and predictability. It helps streamline capacity planning, cost allocation, performance monitoring,

50 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. and automated workload VMware® vCenter™ Server, VIS Delivery Center deployment. It also provides or Citrix® XenCenter™ VIS Self-Service Creator VIS Director a self-service portal, dynamic management consoles, enabling provisioning, and resource administrators to efficiently tiering and pooling. Dell’s open manage large environments Integration Suite architecture provides choice in with a minimal number of VIS Infrastructure what technologies organizations tools—again, leveraging the Advanced Infrastructure Manager select today and in the future. It’s investments already made in the about the ability to connect and infrastructure. Dell services Intelligent hardware integrate multi-vendor server, storage, and networking into a Dell VIS Infrastructure single pool; the ability to migrate As a key part of VIS Infrastructure, from multi-vendor hypervisors Dell Advanced Infrastructure with a simple reboot, and the Manager (AIM) software ability to seamlessly incorporate provides a single management Figure 1. Dell Virtual Integrated System architecture current processes and custom point for physical and virtual applications. resources that helps accelerate The Dell VIS architecture heterogeneous hardware Server and Dell Management consists of three main elements: provisioning while providing Console powered by Altiris™ VIS Infrastructure, VIS Delivery a highly dynamic and flexible from Symantec™. Extending the Center, and the Integration environment. AIM supports AIM advantages include value of virtual Suite (see Figure 1). This design a variety of infrastructure enhanced quality, performance, infrastructures brings together software, elements regardless of make, scalability, and system availability Static IT environments hardware, and services to model, form factor, or access and uptime. It helps increase can lead to major enhance efficiency and enable protocol. Administrators can server utilization and enhances inefficiencies. In this Webinar, learn how Dell rapid IT response to business build a dashboard that provides the ability to perform routine AIM enhances and extends requests. The Integration Suite a holistic view of the data maintenance on both physical virtualization technologies can incorporate existing Dell center, and can invoke many and virtual systems—all while to deliver a truly dynamic data center. management tools into the AIM functions directly from helping to reduce costs Microsoft® System Center common systems management and minimize operational eseminarslive.com/c/a/ Virtualization/Dell091610 Virtual Machine Manager, tools like VMware vCenter interruption. It also helps provide

“In my opinion, the Dell VIS architecture is key to the future of the data center. Moving toward flexible and dynamic data centers is the biggest challenge the IT industry has through the next couple of years, and I believe Dell’s strategy is right on target.” —Gustavo Viceconti CEO of NeuralSoft July 2010

Efficient management for dynamic data centers

This interactive Webcast explores how Dell AIM helps eliminate silos and simplify infrastructure management—enabling IT groups to manage servers, storage, and networks as a single pool.

enterpriseefficiency.com/webinar.asp?webinar_id=29392

the fundamental business storage arrays, Dell/EMC AX4-5 Dell VIS Delivery Center architects to accurately plan continuity and high-availability Series storage arrays, and Dell IT administrators need high for growth and efficiently utilize requirements that facilitate cloud PowerConnect™ switches. IT levels of visibility, automation, resources. or grid computing. staffs need to rack and cable only and customization to help them Dell VIS Delivery Center Achieving high levels of once, and retain the flexibility of a increase productivity, optimize is designed to increase the agility, efficiency, and flexibility multi-vendor approach. performance and resource speed and agility of the data requires the deployment of The use of infrastructure utilization, and increase quality center by delivering enhanced purpose-built servers, storage, management tools helps of service while still controlling levels of visibility and control networking, and fabrics that help increase data center efficiency costs. An understanding over the IT infrastructure. reduce acquisition and operating by integrating IT operations of the relationships and It provides performance costs. Dell offers a range of with broad organizational dependencies between monitoring, capacity planning, infrastructure products to objectives and needs. Dell VIS infrastructure components is service management, support the Dell VIS architecture Infrastructure integrates and critical to identifying and then and automation to help and help meet these needs, automates common tasks remediating performance organizations mitigate the including 11th-generation Dell and workflows, helping free and capacity problems. This impact of resource sharing, PowerEdge™ servers, Dell up resources while enabling intelligence, along with “what-if” and also provides a venue EqualLogic™ PS Series storage dynamic responses to changing scenarios and utilization-based to model cost recovery, arrays, Dell/EMC CX4 Series levels of demand for services. analysis, can help CIOs and IT depreciation, and other financials for IT infrastructure investments. VIS Delivery Center also “We believe we can shorten our deployment compresses the time it takes to deploy applications by time for new applications and servers from standardizing and automating six weeks to just minutes with the help of the way administrators deploy application workloads. With it, Dell Advanced Infrastructure Manager. That’s the IT team can create a self- important in our industry, where we need to service portal for application workloads that provides rapidly meet the growing telecom service templates and monitoring tools, so that the organization needs of consumers, businesses, and carriers can quickly and easily get the around the globe.” most out of its infrastructure investment. —Golan Ben-Oni VIS Delivery Center includes Chief security officer and senior vice president two components: VIS Self- of network architecture at IDT Telecom Service Creator and VIS Director. August 2010 The VIS Self-Service Creator component is an automated

52 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. catalog-based service delivery and management platform that provides IT users with the power “Life cycle management of virtual workloads to deploy and manage their is one of the most important elements of resources while enabling IT staff to respond rapidly to business evolving and optimizing our private clouds. requests and improve control over IT workloads and costs. VIS Self-Service Creator gives us on-demand, VIS Director is the self-service workload provisioning along with command center for the next-generation data center, scalable orchestration to contain sprawl and providing an intuitive, end-to- rightsize capacity with workload leasing end view of the data center that logically links physical and and reclamation.” virtual resources. VIS Director —Matt Brooks gives organizations high levels Senior enterprise architect for Dell IT of visibility and reporting, August 2010 performance and capacity management, and automated remediation. virtualization, followed by nimble, and flexible. CIOs, IT Dell services readiness assessments and architects, and administrators Organizations face multiple technology introductions, or must increase levels of challenges on the path to proofs of concept. This leads standardization and automation a more dynamic cost- and into dynamic data center and develop consistent policies resource-efficient data center. design and implementation, across the organization. To help them deal with these followed by management, Dell VIS provides a path challenges, Dell offers a maintenance, and/or support for the static-to-dynamic broad, customizable range of services. The preliminary phases data center transformation— services designed to simplify include establishing a baseline without sacrificing existing the assessment, design, of projected costs before IT technology choices. Dell’s open implementation, management, decision making, and prioritizing architecture helps ensure that maintenance, and support of initiatives so that they can be organizations retain flexibility IT environments. Dell works implemented on schedule and and choice while gaining with each organization to within budget. Most important, tangible benefits at each step. M. Consuelo Ortiz is a product understand the current IT Dell VIS services are open, By helping simplify data center marketing strategist for Dell. environment and its specific capable, and affordable— management and enhancing challenges. Then Dell maps meaning that organizations can visibility and reporting, VIS out goals and provides incorporate into their plans as enables organizations to practical, action-oriented much or as little as they need. maintain cost-effective recommendations for infrastructures while meeting reducing, removing, or isolating Dynamic data center dynamic business demands. unnecessary complexity. management In delivering a VIS solution, To align with organizational Dell can provide workshops needs and provide the agility Learn more on areas of interest, such needed in today’s fast-paced Dell VIS: as efficient architectures, environment, the data center dell.com/vis cloud computing, and must be cost-effective, dell.com/efficiency

rganizations have always had “hot” Using multitiered information that is accessed more frequently than other data. Ensuring O fast access for this high-demand arrays to optimize information has traditionally required inefficient processes such as making multiple physical application copies to allow several people to view the data at the same time. Today, with large numbers of concurrent users accessing high-demand data performance electronically, ensuring that this data is quickly and easily available to applications and users is an even By Andrew Gilman and Suresh Jasrasaria greater priority. Dell EqualLogic PS6000XVS and EqualLogic PS6010XVS Internet SCSI (iSCSI) storage area network IT organizations have long been challenged to make (SAN) arrays offer an efficient way to meet this need. Incorporating both solid-state drives (SSDs) and sure “hot” data is served to applications quickly and Serial Attached SCSI (SAS) drives in a single array, cost-effectively. Now, they can meet both these these systems can help organizations create a highly challenges with Dell™ EqualLogic™ hybrid storage responsive yet cost-effective platform for a range of multitiered workloads, including virtual desktop arrays that use multiple tiers in a single enclosure. infrastructure (VDI) with VMware View as well as business-critical database applications.

Addressing the high-demand data challenge Rapid access to high-demand data is essential to the performance of mission-critical databases and key business applications when deployed in a physical or virtual environment. If an application is unable to retrieve high-demand data from storage quickly enough, the user experience for these applications can suffer. This problem can be especially acute for applications that serve multiple users. Activities such as month-end database processing can Data on demand exacerbate the problem by greatly multiplying Dell EqualLogic PS6000XVS and EqualLogic simultaneous demand for key data. Adding to PS6010XVS arrays include SSD and SAS drives in the problem is the pace at which data is a single enclosure and provide a cost-effective, growing, increasing the pressure on storage highly efficient approach to ensuring fast access system performance. to high-demand data. Networked storage and extensive use of SSDs help deliver fast access to data and increased • Multiple storage tiers with SSDs and SAS application performance—but these technologies drives in a single enclosure can be complex, as well as cost prohibitive for all • Automated load balancing for an appropriate but the largest of organizations. As data centers balance of performance and cost become increasingly consolidated, providing • Integration with the VMware® vSphere™ 4.1 application performance on demand while hypervisor to increase responsiveness in keeping costs low has challenged even the most virtualized environments sophisticated IT organizations.

54 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Overcoming limitations of traditional multitiered load balancing When using traditional load balancing across Turning up the speed multiple arrays, administrators must be cautious Dell testing shows that using EqualLogic firmware version 5.0 in about migrating data to a lower tier of storage. conjunction with VMware vStorage application programming interfaces For example, the latency involved in moving a (APIs) can help significantly accelerate VMware vSphere 4.1 performance. data volume to a lower-tier array can reduce The firmware is designed to increase the responsiveness of VMware application performance. When high and low virtual infrastructures running on EqualLogic arrays and offers a variety of data access frequency coexist on the same advantages, including the following: volume or file, it is often difficult to manually separate frequently accessed data from Reduce time to complete provisioning by up to 72 percent infrequently accessed data. Moreover, when storage systems perform volume- or file-based Reduce SAN network traffic by up to 95 percent when using auto-tiering based on actual access frequency, Full Copy offload volumes or files may need to move frequently between tiers of storage, causing thrashing. Reduce processor utilization by up to 75 percent when using A cost-effective approach to reducing access Full Copy offload* time for frequently used data is to enable content- *All three results based on testing performed by Dell Labs in May 2010 comparing EqualLogic firmware based auto-tiering between high-performance version 4.3 with EqualLogic firmware version 5.0 Actual performance will vary based on configuration, usage, and workload variability. SSDs and less-costly SAS drives within the same array. Dell EqualLogic PS6000XVS and EqualLogic PS6010XVS iSCSI SAN arrays follow this strategy, providing an efficient way to implement simplified cabling and exceptional sequential multitiered load balancing. performance by using 10 Gigabit Ethernet (10GbE) interfaces. Dell EqualLogic PS6000XVS and EqualLogic PS6010XVS array capabilities Integration with VMware virtualization The high-performance Dell EqualLogic Dell EqualLogic PS6000XVS and EqualLogic PS6000XVS and EqualLogic PS6010XVS arrays PS6010XVS arrays are designed to boost data are designed to support fast access to high- efficiency for tiered workloads by automatically demand data in a multitiered volume. Each array migrating data to the appropriate drives and includes eight low-latency SSDs and eight cost- by leveraging tight integration with VMware efficient 15,000 rpm SAS drives for multiple tiers virtualization technologies, including of storage within a single array. VMware vStorage application programming On-board intelligence provides automatic interfaces (APIs). load balancing across these drives, EqualLogic firmware version 5.0, included helping deliver the appropriate balance of with all EqualLogic arrays, is closely integrated responsiveness and cost for tiered workloads. with VMware vStorage APIs. The firmware allows EqualLogic firmware enables efficient load VMware vSphere virtualization software running balancing between the SSD and SAS tiers. on servers with the VMware ESX 4.1 hypervisor Intelligent data tiering and automated to offload storage-related functions from the management are built into both arrays, hypervisor layer to the EqualLogic firmware. along with snapshots and replication for data This offload capability helps increase the protection and disaster recovery capabilities, responsiveness of VMware virtual infrastructures requiring no additional licensing costs.1 The running on EqualLogic arrays, and enables EqualLogic PS6010XVS array also offers intelligent data tiering for multitiered workloads

1 For more information on intelligent automated tiering in EqualLogic storage arrays, see “Cost-effective tiering for virtualized iSCSI SANs,” by Achmad Chadran and Keith Swindell, in Dell Power Solutions, 2010 Issue 3, dell.com/downloads/global/power/ps3q10-20100493-chadran.pdf.

Accelerating such as virtual remote desktops the linked-clone capability in VMware cloud services running in VMware vSphere virtual View Manager with View Composer When SAP asked infrastructure-as-a-service (IaaS) environments. (For examples, see to allow multiple desktop images to provider Xtium to deliver “test drive” environments the “Turning up the speed” and work from the same gold template for prospective customers, the company “Accelerating cloud services” sidebars image, reducing the need for spinning quickly took advantage of EqualLogic firmware in this article.) up new whole desktops and avoiding version 5.0. The thin clone technology and tight this situation. VMware integration have provided Xtium with a Responding intelligently robust, scalable, easy-to-manage platform that to data requirements Understanding how The ability of the Dell EqualLogic enables quick, cost-effective provisioning while Dell EqualLogic arrays PS6000XVS and EqualLogic enable efficient cloning supporting aggressive service-level agreements. PS6010XVS arrays to sense and Dell EqualLogic PS6000XVS and The results have been dramatic: intelligently respond to data EqualLogic PS6010XVS iSCSI SAN requirements is applicable to arrays use a combination of intelligent 30 times faster provisioning of clones for multiple use cases, including data placement and SSD acceleration test drive environments (from one hour to VMware View–based VDI and to automatically deliver performance two minutes) business-critical applications such where and when it is needed. To use as Oracle® databases and SAP® the VDI boot-up example, the data 5 times faster copies using Xcopy for single implementations. In each of these optimization process includes storing instances of test drives and virtual desktops use cases, EqualLogic PS6000XVS the gold master image on the top-tier and EqualLogic PS6010XVS iSCSI SAN SSDs, providing low-latency data 90 percent savings on disk space using arrays help organizations increase access and accelerating the cloning thin clones the efficiency of data processing of the image as it is booted up using 70 percent more hosted virtual machines operations. the linked-clone capability in VMware on fewer physical resources with atomic VDI offers one example. With View Composer. locking potentially hundreds or thousands The gold master image can be of desktop users booting from the cloned rapidly to start virtual desktops 50 percent fewer volumes to manage same file, a VDI implementation while minimizing the use of disk presents significant challenges for space. When users are ready to shut The firmware upgrade came at no serving hot data. For example, an the desktops down, each user’s additional cost to Xtium, and the combination organization might have a thousand data changes can be saved and all of technologies like thin provisioning and thin call-center workers all coming to of the clones deleted, so that the clones is allowing the company to deliver work at the same time each morning; organization is still only storing one new and enhanced services going forward. when using virtual desktops, these gold master copy at the end of the “Customers are enjoying services that are employees would all turn on their day. Infrequently accessed user and making them more efficient and providing new clients, access the same file, and application data can be moved to the revenue,” says Xtium CTO Tim Vogel. “And the boot up the same gold master image SAS portion of the array for cost- new EqualLogic firmware enables us to grow at the same time. effective data storage. dynamically, so it’s a win-win.”* This scenario could easily lead The linked-clone capability in to a daily “boot storm,” resulting VMware View Manager with View

*For the full story, see “Xtium provisions tier-1 apps and virtual in resource contention, delayed Composer enables virtual desktop desktops 30 times faster with Dell EqualLogic and VMware vSphere 4.1,” by Dell Inc., August 2010, dell.com/downloads/global/ booting, and sluggish performance administrators to clone and deploy casestudies/2010-xtium-10008801.pdf. that can substantially reduce user multiple desktops from a single productivity. EqualLogic PS6000XVS centralized VM called a master VM and EqualLogic PS6010XVS arrays use (see Figure 1). The clones do not

Linked clone 1

VMware View Composer System disk Figure 1. Linked cloning in VMware View (250 MB) Manager with View Composer User data disk Replica (10 GB) Linked clone 2 Gold master VM (10 GB)

System disk (250 MB) User data disk

Linked clone n

have any link to the master VM, which can even Increasing the efficiency of data be deleted without affecting the clones. The processing operations first time a desktop clone is created, a uniquely By helping increase the efficiency of data identified copy of the master VM—called a processing operations, including application replica—is also created. Desktop clones are access to high-demand data, Dell EqualLogic anchored directly to the replica, and changes to PS6000XVS and EqualLogic PS6010XVS iSCSI SAN the master VM do not affect the anchored clones. arrays can support growing data requirements Because the clones in this environment are without a corresponding increase in cost and connected to a common source, View Composer complexity. And by incorporating intelligent enables the centralized management of desktops data tiering, automated management, and other while maintaining a seamless user experience. features, these arrays provide a high-performance, cost-effective way to meet the needs of today’s Creating an end-to-end virtual critical multitiered workloads. desktop platform

Together, Dell and VMware offer an end-to-end Andrew Gilman is a solutions marketing virtual desktop platform using Dell EqualLogic manager on the Dell Large Enterprise Storage PS6000XVS and EqualLogic PS6010XVS iSCSI Solutions team responsible for virtualization marketing activities. SAN arrays, Dell PowerEdge™ servers, and Dell PowerConnect™ switches along with VMware Suresh Jasrasaria is a product marketing View and vSphere—an approach that enables senior consultant in the Dell Enterprise Storage Product Group, and has worked in the data organizations to standardize on a common storage industry for more than 20 years. platform and a tested, scalable architecture. The Dell Virtual Remote Desktop (VRD) architecture is Learn more designed to provide a cost-effective, automated solution for managing client platforms. Using Dell EqualLogic PS Series: Dell VRD, organizations can take advantage of dell.com/psseries existing investments in VMware virtualization and dell.com/equallogic extend powerful cloud delivery features to the VMware virtualization: desktop infrastructure.2 vmware.com/virtualization

2 For more information on the Dell VRD solution, visit dell.com/fcs.

How desktop virtualization enhances end-user productivity

By Todd Mitchell and Doug Coombs

By emphasizing customized, secure workspaces for end users, Symantec™ Endpoint Virtualization Suite helps to maximize productivity while simplifying management and security for IT administrators.

s traditional lines between virtualized applications on top of the organization’s virtualization, management, and existing OS distribution method—regardless of the security continue to blur, desktop underlying compute environment, connectivity, A virtualization is emerging as a and location. The result offers consistently available, compelling option for many organizations. highly productive workspaces across a wide Interestingly, leaders in this field commonly variety of infrastructures, including VMware® and discuss management and security benefits Citrix® platforms as well as traditional thick clients more than they do the underlying abstraction (see Figure 1). Furthermore, SEV uses a single technologies they implement. Therefore, when management framework to both deliver dynamic evaluating desktop virtualization options, it’s desktops and enable proactive license compliance. a best practice to keep in mind that vendors Symantec’s heterogeneous cross-platform, take vastly different architectural approaches to cross-model approach extends the many Symantec solving similar problems. Architectural differences benefits of desktop virtualization across as Endpoint emphasize value at different points in the overall much of the environment as possible, and Virtualization solution stack, and can dramatically limit the turns the challenge of proprietary system Suite in action degree to which those management and security lock-in into an opportunity for optimization and

Visit Symantec online benefits are realized in the real world. standardization. This approach offers simplified, to view detailed video Because end users are key to creating centralized management as well as the freedom demonstrations and to learn more about Symantec business value, Symantec approaches the virtual to move users across a variety of compute Endpoint Virtualization Suite. desktop with an exclusive focus on end-user models, hypervisors, and vendors as needs and

dell.symantec.com/psq productivity. By focusing on the workspace rather budgets evolve. than on the hypervisor, and providing each user In addition to this hypervisor-agnostic with a secure shell customized to meet individual approach for simplified management across needs, Symantec Endpoint Virtualization Suite both virtual and thick clients, SEV offers many (SEV) helps to maximize end-user productivity additional advantages: while offering IT administrators simplified, flexible control over the environment. • Integrated, real-time license management: SEV is designed to provision applications to Delivering robust workspaces to end users users or devices in real time. Because software Rather than delivering a hypervisor or an OS, is not moved to the system before it is used, a SEV layers together a dynamic user interface license is provisioned only after the software based on combinations of local, streamed, and is actually launched, helping prevent the

58 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Shared Dedicated Proﬁles users as well, by allowing virtualized applications to continue desktops desktops Applications and data behaving as expected: functions such as copy and paste and file associations remain in place without additional up-front packaging work. • Optional secure shell: Symantec dynamic workspaces can present either the entire Windows shell or a locked-down, customized shell showing only specific interface elements or applications. • Follow-me printing: As users move between systems, they can automatically print to a local printer. • Fast login times: Symantec software can help dramatically increase user productivity in challenging multiuser environments

Profile and data Profile and data Profile and data such as hospitals. • Remote application reset without reinstallation: Some Applications Applications Applications products require that broken applications be uninstalled

Desktop Desktop Desktop and reinstalled, sometimes forcing an IT engineer to visit the endpoint. SEV can quickly, easily, and remotely reset applications to a known good state. • Mix-and-match installation: Unlike some solutions, SEV does not require installing all features and components at once— IT administrators can implement only the parts they need.

Figure 1. Symantec Endpoint Virtualization Suite provides end users with a Administrators can choose to stream applications, virtualize dynamic, productive workspace regardless of location applications, or both.

Maximizing productivity while simplifying management purchase of unneeded software. For applications that allow Desktop virtualization promises numerous benefits, but can also add license reclamation, SEV enables administrators to set complexity. SEV allows IT administrators to provide end users with rules for harvesting unused licenses and assigning them to optimized performance while simplifying management and security other users. Simply using this single feature across a few functions. Whether by enhancing end-user productivity through high-priced applications can provide a compelling return increased mobility, supporting system security through compatibility on investment. with existing applications, or avoiding unnecessary costs through • Fast, heterogeneous application packaging: Symantec automation, SEV can help both end users and IT administrators application packages are portable between 32-bit Microsoft® make efficient use of available computing resources, enhancing Windows® XP, 32- and 64-bit Windows Vista®, and 32- and productivity in both spheres. 64-bit Windows 7 environments—enabling administrators to truly “package once, run anywhere” in both client- and server-based Todd Mitchell is a technical director in the Global Strategic computing environments. Alliances organization at Symantec. • Enterprise cloud ready: Symantec’s lightweight architecture Doug Coombs is senior director of product management for helps increase performance for applications that require endpoint virtualization at Symantec. streaming across low-bandwidth lines. • Application visibility: The highly proprietary nature of some application virtualization platforms means that those platforms Learn more are unlikely to be supported by existing management and security tools—often requiring the purchase of replacement tools Symantec Endpoint Virtualization Suite: symantec.com/business/endpoint-virtualization-suite from the desktop virtualization vendor. Symantec virtualization technologies are designed not to abstract applications from the Dell and Symantec: OS so much that they become lost to management and security dell.com/symantec tools. The added application visibility provided by SEV benefits symantec.com/dell

uch has been said about the role of cloud computing in the future of IT. Public, private, hybrid, and community M clouds merge together into a utopian future of seamless, unending capacity and unparalleled customer service. Although such a goal helps focus activity, IT leaders today recognize that this end is easier described than built. Still, in the short term, there are real and tangible cost and flexibility benefits that can be reached with cloud computing.

Defining the cloud A brief definition helps frame the challenge: a cloud is a pool of resources from which users can request a service to be automatically and rapidly provisioned for their needs. The abstraction layer of cloud management obscures the location, hardware, and sometimes even size of that resource from end users, providing the service without the burden of managing Successful liftoff: systems. Meanwhile, on the back end, a cloud should be intelligent enough to optimize the use Achieving cloud of the underlying hardware for cost savings, flexible enough to reallocate resources to meet service-level agreements, and (in some cases) elastic enough to computing in the seek resources from third-party clouds when needed. That’s a lot to ask from an IT environment that, enterprise until very recently, sat in one spot and ran a single OS and workload day in and day out. Of course, there are By Lilac Schoenbeck a whole range of IT components that come into play with cloud computing. Dell offers different solutions that facilitate the Deploying BMC® Cloud Lifecycle Management on a data center transformation and help IT organizations build optimized data centers and cloud computing ™ foundation of Dell servers, storage, and switches can environments. For mainstream customers, Dell offers help organizations create dynamic cloud computing the Virtual Integrated System, an architecture that environments designed for flexible, on-demand enables rapid resource deployment and on-demand delivery and management of services. And for large service provisioning and simplified management. enterprise customers with more demanding IT challenges, Dell has partnered with BMC to provide dynamic cloud management solutions using BMC Cloud Lifecycle Management deployed on Dell PowerEdge™ servers, Dell EqualLogic™ storage, and Dell PowerConnect™ switches. A discussion of the cloud necessarily begins with the servers: without scalable processing power, there is no cloud. Dell PowerEdge servers provide a scalable, manageable, and flexible foundation to the cloud architecture.

60 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. The servers must in turn be augmented by Controlling the cloud life cycle storage, because the data and virtual machines Cited for its business service management in the cloud must be stored somewhere. leadership,1 BMC Software provides a Storage is often an afterthought of the cloud comprehensive and unified platform designed to architecture, which can be dangerous given that simultaneously optimize IT costs, demonstrate the portability of the workload is often strongly transparency, increase business value, control tied to the portability of the data. Provisioning risk, and ensure quality of service. BMC a cloud service requires provisioning storage brings together the benefits of traditional IT Unified alongside it, and ensuring that this storage is as management—including operational excellence, management well managed as the services it supports. The automation, and service delivery models—and In this video, BMC CIO Mark simplified storage management and seamless merges them with the dynamic potential of Settle discusses why unified expansion features of Dell EqualLogic PS Series cloud architectures. management is critical to securing the benefits of cloud Internet SCSI (iSCSI) storage area network (SAN) BMC Cloud Lifecycle Management knits computing. arrays enable scalable performance and reliability together the various cloud component resources bmc.com/videos/94779114.html as the cloud grows. and provides the foundation for a strong, flexible, Networks are the connective tissue of and valuable cloud infrastructure that supports a cloud. Often a bottleneck, bandwidth to IT operations and delivers exceptional service the cloud and between elements in the quality (see Figure 1). It provides an operational cloud must be allocated and prioritized so model for the life cycle of private cloud resources that it doesn’t become a constraint on the and the utilization of public clouds in a hybrid flexibility and elasticity of the environment. model. Each resource in the environment goes Allocated network resources are often also through a life cycle that, when defined and the location of security measures, so the appropriately automated, provides a seamless and proper translation of isolation requirements predictable model for both IT departments and into the cloud architecture is critical to the organization as a whole. helping ensure that security is maintained or enhanced as workloads move to the cloud. Dell PowerConnect managed switches provide End Service request user a high-performance and secure foundation Self-service for networking the components in the cloud portal infrastructure. On top of that foundation, Service BMC BladeLogic® Network Automation catalog enables dynamic configuration change to Service request management Public VM VM VM meet the changing needs of the cloud. (automated) cloud Conﬁguration Once the physical components have come management Virtual system (CMS) machines together, additional components layer on or CMDB

top: hypervisors, operating systems for each Service Orchestration cloud service, and the stack of middleware and retirement and provisioning Network applications that comprise the service most tangible to the customer or internal group. For users of the service, the cloud need only Physical Storage servers display the highest layers of the stack. For IT Applications departments, however, the full complement of pieces must come together to deliver the value of Operations and governance cloud efficiently and economically. Performance Compliance Financial management management management

1 “The IT Management Software Megavendors,” by Jean-Pierre Garbani and Peter O’Neill, Forrester Research, Inc., August 12, 2009. Figure 1. BMC Cloud Lifecycle Management operational model

Customizable Server instances dashboard for filter events and alerts

Figure 2. BMC Cloud Lifecycle Management Existing user self-service portal instances

Request for new server instance Customizable “quick pick” list

Server instance details

The platform comprises six key and the appearance of the organization— administrators to maintain tight control functions: a self-service portal, a service helping dramatically simplify the end over the cloud environment and the catalog, flexible provisioning, closed-loop user’s interaction with the cloud. As users services deployed. compliance, decommissioning, and hybrid become increasingly sophisticated (and Based on the BMC Atrium® cloud operations. service offerings become more complex), configuration management database the self-service portal can continue (CMDB), the service catalog not only Self-service portal to address the needs of research and offers new service instances that users The most visible user-facing part of the development, software development, can select from the portal; it also cloud computing environment is the application deployment, and even internal represents the authoritative record of the self-service portal (see Figure 2). Through business customers. functional components that make up that the intuitive Web browser–based interface service, from servers and storage to OS of BMC Cloud Lifecycle Management, Service catalog and application binaries and licenses to end users can manage the services they Behind the self-service portal lies a service the IT resources needed to maintain the have requested from the cloud, including catalog, which acts as a listing of the service in production. For example, as IT turning them on or off and requesting services and options available to users of staff install new servers and storage and additional time or resources. Built on the cloud. Ranging from resource profiles to publish those resources to the catalog, the BMC Remedy® Action Request operating systems to full listings of available they are automatically factored into the System™ platform, the self-service portal applications, this catalog controls what model of capabilities that may be made is designed to be inherently multi-tenant, appears to each user in the self-service available to users. scalable, and secure. portal. Behind the scenes, the service From this portal, users can also request catalog also captures the intelligence and Flexible provisioning new services through a simple wizard that processes to be followed to implement the To help maximize the flexibility of the exposes specific options that are available operating systems and applications in an service stacks for users, BMC Cloud to that user based on his or her role in the automated fashion. Lifecycle Management supports a versatile organization. Options range from different The service catalog is a major source underlying provisioning capability. resource sizes, service tiers, and operating of flexibility and control in the cloud Traditional virtualization provisioning is systems through application stacks and environment. The many combinations image based, requiring IT staff to either higher-level services, such as compliance available from the service catalog represent restrictively standardize on a very small set and monitoring. a diverse and vast set of configurations to of images or, alternatively, manage a library The portal provides an approachable help meet a user’s specific needs. But the of hundreds of unique images. The BMC front end to the cloud environment, constraints, role-based access, and policies Cloud Lifecycle Management approach is customized for both the functionality reflected in the catalog enable the cloud one of controlled customization, delivering

62 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. flexibility for end users within the constraints the physical to the virtual and even to the are proliferating, offering an array of designed by IT departments. public cloud. It stands to reason, then, that options for enterprises looking to extend Beginning with the physical resources and clouds should be managed through the the scale of their IT operations—whether the OS and climbing up the stack through same rigorous approaches that have grown temporarily or permanently—with external middleware and applications, the provisioning from years of investment and experience resources. Critical to using those resources, capability can even layer compliance rules in IT management. The CMDB should however, is a management model that can and monitoring tools into each service track cloud services just as it tracks other integrate the services seamlessly into the delivered. The physical foundations—Dell items. And perhaps most important, change organization’s environment. PowerEdge servers—can be provisioned management and compliance should be BMC Cloud Lifecycle Management rapidly and securely using BMC BladeLogic applied just as thoroughly in the cloud as on enables IT departments to take advantage of Server Automation integration with the Dell other systems. public cloud resources, like those of Amazon’s Lifecycle Controller. In cloud environments, As dynamic as the cloud environment Elastic Compute Cloud (EC2) service, from this integration enables the newly provisioned is, change management can be challenging within the same self-service portal as the servers to be inventoried and accounted for IT administrators. Inventories and private cloud resources. Regardless of for in the service catalog before the server lists are difficult to maintain in a world of whether the ultimate destination of their is even powered up. Upcoming integration shifting workloads and newly provisioned cloud service is made clear to end users, with Dell Advanced Infrastructure Manager services. Closed-loop compliance— or whether provisioning decisions are (AIM) is expected to provide a unified, logical for both configuration and regulatory automated on the back end, this capability abstraction of the disparate components reasons—helps ensure that a change is enables great flexibility for enterprises in the cloud infrastructure, so they can be identified, a remediation is enacted, and seeking to efficiently meet demand. organized into flexible resource pools to be the result is confirmed. This automation is dynamically allocated, and re-allocated, to accomplished through BMC BladeLogic Building robust, flexible meet the desired availability and capacity Server Automation Suite, which integrates environments metrics of the services offered by the cloud. with the Dell Lifecycle Controller in With comprehensive life cycle automation These resource pools represent the core PowerEdge servers. and customizable components, building blocks of the cloud. BMC Cloud Lifecycle Management BMC BladeLogic Server Automation Decommissioning enables IT organizations to build cloud manages change in the cloud. As new Cloud services are designed to be transient. environments that are robust and flexible. capacity is requested, BMC BladeLogic Easy to request and provision, they can Leveraging underlying resources from Server Automation works with Dell AIM be instantiated for hours or days and then Dell PowerEdge servers, Dell EqualLogic to draw from the resource pools and to retired, freeing the resources for the next iSCSI SAN arrays, and Dell PowerConnect instantiate virtual machines, provision user. The retirement of resources is a critical switches, an enterprise cloud can help to applications, and help ensure that the part of that cycle. significantly lower administrative burdens, appropriate security, compliance, and BMC Cloud Lifecycle Management reduce costs, and ultimately deliver the governance constraints are enforced automates this step on a calendar and helps greatest value to the business: excellent on an ongoing basis within these virtual ensure that cloud services aren’t simply customer satisfaction. machines. Backed by the robust service neglected in the environment, taking up catalog, automation capability, and role- costly resources and delivering no value. Lilac Schoenbeck is senior manager of based access controls, this provisioning Instead, on a schedule, cloud users are product marketing for cloud computing approach helps give users precisely the prompted to decommission or extend their at BMC Software. stack they require while maintaining the services, helping ensure that the life cycle tight controls necessary to manage a continues seamlessly. Learn more complex IT environment. Hybrid cloud operations Dell cloud computing: dell.com/cloud Closed-loop compliance One of the primary values of the cloud Clouds are not islands in the data center. architecture is its elasticity and scalability, Rather, a cloud should be a component even beyond the organization’s data BMC cloud computing: of a broad array of infrastructure, from center. Public cloud service providers bmc.com/cloud

How end-to-end services can transform Oracle environments

By Tim Daigle and Thomas Kopec

The Dell Services Oracle Practice provides end-to-end services to organizations running Oracle® software—including applications, middleware, and database platforms—to help drive cost and complexity out of the IT infrastructure.

ell’s acquisition of Perot Systems in provide end-to-end services for a comprehensive 2009 significantly expanded the Dell range of Oracle technologies. Services portfolio. Perot Systems was The Dell Services Oracle Practice is charged Da pioneer in the field and a world-class with supporting and extending the ongoing efforts provider of applications, technology, infrastructure, of both Dell and Oracle to take cost and complexity and consulting services—and the addition of its out of the IT infrastructure. Many companies spend capabilities has enabled Dell Services to expand and a majority of their IT budgets simply on maintaining enhance the services it can provide for businesses the status quo—just keeping the lights on, from an and other organizations. IT perspective. Dell’s goal is to help organizations A key component of this broadened approach drive that number down to under 50 percent, so is Oracle-related services. Perot Systems had a that they can instead focus on new and strategic consulting practice focusing on Oracle applications initiatives. Spanning strategy, design, implementation, in place since 1995, while Dell has built up a range maintenance, and outsourcing, the Dell Services of services based on running Oracle middleware Oracle Practice opens up a wide range of options to and database software on Dell™ servers and help organizations meet these goals—enabling them storage. Bringing these capabilities together, Dell to simplify their operations, reduce costs, and get the has created the Dell Services Oracle Practice to most out of their Oracle platforms and applications.

64 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Providing comprehensive services The Dell Services Oracle Practice includes some 1,100 professionals Accelerating return and a network of Oracle-specific centers in Brazil, China, India, on investment Mexico, and the United States. Dell is a Specialized Oracle Platinum Partner, the highest level in Oracle’s partner programs. And Dell Reducing IT complexity can pay off in Services brings extensive expertise in Oracle applications, middleware, increased productivity and lowered costs. and databases, supported by a cost-effective global delivery model. The expanded Dell Services portfolio The Dell Services Oracle Practice works with a wide range of Oracle includes a variety of targeted services technologies. It offers services focusing on applications, including designed to standardize, simplify, and Oracle E-Business Suite, PeopleSoft®, JD Edwards®, Hyperion, Oracle automate ongoing operations. Business Intelligence Enterprise Edition, Oracle Communications Billing and Revenue Management, and Siebel software. The practice Platform migration: Dell Services can also works with Oracle Fusion middleware and service-oriented help organizations assess and carry out architecture software, as well as Oracle’s industry-specific solutions for migrations from proprietary systems manufacturing, travel and transportation, consumer packaged goods, life to Linux® OS–based Oracle and Dell science and pharmaceuticals, education, health care, and government. platforms, including moving applications To help organizations get the most out of these technologies, and database systems to the new platforms the Dell Services Oracle Practice provides consulting that focuses and implementing Oracle Real Application on identifying opportunities for transformation or improvement Clusters—enabling organizations to take through customization, modernization, and maintenance. And advantage of cost-effective, standards- to help take them advantage of those opportunities, the practice based infrastructures. provides development, customization, implementation, integration, testing, and maintenance services. Preconfigured software: Options include Meanwhile, Dell Services also helps organizations simplify data preconfigured analytics for the health care, center operations with assessment, design, and implementation manufacturing, consumer, and packaged services that focus on Oracle Database, including Real Application goods industries, as well as supply chain Clusters, in a standards-based x86 environment. These services— and human resources applications. designed to enhance data availability while lowering total cost Preconfigured industry solutions are also of ownership—are based on best practices developed through available for milled products, education, and rigorous joint engineering between Dell and Oracle. Organizations state and local government. can also offload database administration and get customized support through Dell Services. (For additional examples, see the Accelerators: Dell Services can help speed “Accelerating return on investment” sidebar in this article.) Oracle E-Business Suite implementations with a methodology that uses Oracle Addressing key challenges Business Accelerators to quickly provide In addition to broad Oracle-specific capabilities, the Dell Services Oracle scalable, industry-specific functionality. Practice can help organizations meet a range of specific challenges: These accelerators are available in prepackaged, fixed-price, and fixed-scope • Business transformation: Competition in today’s global application bundles. economy can be brutal. Dell Services uses best practices and organizational change management in conjunction Cloud services: Services can include cloud with Oracle enterprise resource planning (ERP) and supply chain management (SCM) software to help transform assessments, on-demand Oracle cloud businesses as they expand into new markets and products. development environments for upgrades • Application management: Using a service-level agreement and add-ons, and a JumpStart Cloud service (SLA)–based delivery model, Dell Services focuses on delivering that enables accelerated enterprise resource both cost-effective, reliable service and ongoing quality and planning (ERP) implementations. productivity enhancements. Delivery is based on IT Infrastructure

Library (ITIL), Capability Maturity Model Integration its onshore and offshore resources to help deliver (CMMI), and Application Implementation predictable benefits and return-on-investment Methodology (AIM) techniques along with a mix metrics in ERP upgrade efforts. This service of onshore, nearshore, and offshore resources helps organizations simplify current software customized to help meet each organization’s approaches; reduce or eliminate customization; specific needs. Dell application management streamline corporate portal services such services include functional and technical support, as financial, payroll, human resources, and support for ancillary systems, core database customer relationship applications; and set administration services, infrastructure hosting, the stage for significant operational savings. network services, and help-desk support. • Governance, risk, and compliance (GRC): • Business intelligence: Dell helps organizations Organizations must manage growing amounts draw on information from multiple sources of data for financial reporting, data security, to provide management with comprehensive records retention, risk reporting, and more. reporting on business performance, along The Dell Services Oracle Practice helps identify with deep-dive analytics and dashboards to and involve key leaders and stakeholders enhance decision making. Services include the throughout the organization, and brings Dell Services Optimized Business Intelligence them together to develop an understanding program, an approach that integrates key of what is needed to build a case, ascertain financial and operational analytics throughout return on investment, and develop a GRC the organization with a focus on using best program plan based on Oracle solutions. practices to standardize processes and drive profitability. Dell also provides data models for Reducing IT cost and complexity key industries, and an optimization program The creation of the Dell Services Oracle Practice designed to jump-start business intelligence builds on the long and successful relationship deployments. As a starting point, a complimentary between Dell and Oracle—a collaboration that Discovery Session involves key stakeholders and can offer significant benefits to organizations Dell professionals to help identify the potential running Oracle software on Dell hardware. value of a given business intelligence initiative. Ranging from high-end strategy work and global • ERP evaluation and enhancement: Many factors systems design to implementations and upgrades, can affect an enterprise system’s ability to meet multiyear support, and application outsourcing, the needs of the organization: applications the services offered by this practice can help may have been implemented incorrectly, there organizations gain the knowledge and tools to may have been inadequate attention paid help simplify the data center, increase productivity, to change management and training during and keep IT aligned with strategic goals. implementation, certain features and functions may be underutilized, and requirements and the Tim Daigle is the global Oracle Practice organization itself can change over time. The leader at Dell. ERP Health Check service gauges how effectively an organization is using its current Oracle Thomas Kopec is a senior technical account manager with Oracle. applications, and then helps it realize the full value of those applications. The service addresses problems by evaluating the existing deployment, Learn more mapping requirements to the application configuration, identifying performance gaps, Dell Services: dell.com/services and providing a road map for filling those gaps. • ERP upgrades: The ERP Upgrade service leverages Dell and Oracle: Dell’s Oracle applications knowledge; its cost- dell.com/oracle effective tools, templates, and methodologies; and oracle.com/dell

Intuitive alphanumeric display: Energy savings: Get detailed UPS and power quality A patent-pending “green” mode achieves information at a glance – including status, online efficiencies approaching 99 percent, about, and diagnostic log menus in your reducing heat loss and utility costs. choice of up to five languages.

Configurable interface: Set up and control key UPS parameters and functions using the intuitive navigation keys. On rack/tower convertible models, the display rotates 90 degrees for easy viewing.

Contact your Dell Representative today and ask for APC Smart-UPS

If you want Legendary Reliability inside, it had better say APC outside. What do you get when you combine 25 years of Legendary Reliability with the latest in UPS technology? Introducing the new APC Smart-UPS range of interactive, intuitive, and energy-saving UPSs, designed to protect critical server and network equipment from power threats and downtime. New APC Smart-UPS: Smarter. Easier. Greener. Thanks to millions of dollars in research, APC can proudly claim that only the new Smart-UPS features the unique battery life expectancy predictor, telling you the exact month and year for battery replacement. Precision temperature-compensated charging extends battery life; unique power meter function monitors energy usage; and a patent-pending “green” mode boosts online efficiencies up to 99 percent, saving on utility costs. Plus, the Only APC offers the most interactive LCD provides detailed status, configuration, and diagnostic information previously available only via technologically advanced, software. user-friendly features, and the guaranteed reliability you need When dollars count and performance is critical, insist on the more intelligent, more intuitive APC Smart-UPS. to protect your critical data and Now more than ever, the name on the outside guarantees reliability on the inside: APC Smart-UPS. equipment. Look for APC on the outside to ensure Legendary Reliability on the inside.

Download a FREE copy of APC White Paper #10, “Preventing Data Corruption in the Event of an Extended Power Outage.” Visit www.apc.com/promo Key Code t441w Call 888-289-APCC x0000 • Fax 401-788-2797

©2010 Schneider Electric Industries SAS, All Rights Reserved. Schneider Electric, APC, Smart-UPS, and Legendary Reliability are owned by Schneider Electric, or its affiliated companies in the United States and other countries. e-mail: [email protected] • 132 Fairgrounds Road, West Kingston, RI 02892 USA • 998-2158

998-2158_DellComputer.indd 1 5/14/2010 1:43:24 PM Networking

Enhancing Ethernet network controllers for server virtualization

By Dhiraj Sehgal, Abhijit Aswath, and Srinivas Thodati

Broadcom and Dell are working with a range of virtualization providers to enhance Broadcom® network controllers in Dell™ PowerEdge™ servers to help reduce the bottlenecks and overhead that can impair performance in virtualized environments.

n enterprise IT environments, virtualization offers of Dell PowerEdge servers as LAN on Motherboards numerous advantages—enabling organizations (LOMs), mezzanine cards, and standard network to consolidate and share computing resources, interface cards (NICs), and are used as part of I increase control and efficiency, and operate their Dell’s overall virtualization solutions. Broadcom is data centers cost-effectively. Getting the most out currently working in a two-phase effort to enhance of a virtualized environment, however, requires more its controllers for use in virtualized environments, than simply deploying virtual machines (VMs). An helping organizations using Dell systems to maximize effective networking infrastructure is essential. the benefits of their virtualization initiatives. Like other hardware resources in Dell systems, Broadcom network adapters are virtualized to the Phase 1: Removing bottlenecks VM. Virtualization platforms use a hypervisor-based and increasing performance architecture—also known as a VM monitor (VMM) In the first phase of enhancements, Broadcom architecture—that hides the physical characteristics of worked to remove virtualization bottlenecks the computing platform and allows unmodified VMs to and increase system performance by providing run concurrently on host platforms. a number of additional features. For example, This design means that virtualization comes at the Broadcom Ethernet network controllers now support cost of reduced performance. Today’s virtualization stateless offloads such as TCP checksum offload, architectures include VMs with device drivers, I/O stack, which enables network adapters to compute TCP and applications running on top of a virtualization layer checksums on transmit and receive, and TCP large that includes device emulation, I/O stacks, and physical send offload, which allows the TCP layer to build a device drivers that manage the Ethernet network TCP message up to 64 KB long and send it in one call controllers. This virtualization layer adds overhead down the stack through IP and the Ethernet device and degrades system performance, driving up driver, saving the host processor from having to processor utilization and reducing available bandwidth. compute the checksum in a virtualized environment. Broadcom, a networking leader in Gigabit Ethernet In addition, Broadcom’s jumbo frame support (GbE) and 10 Gigabit Ethernet (10GbE) network in virtualized environments is designed to save controllers, has been working closely with a variety processor utilization by reducing interruptions, and of virtualization vendors—including VMware, to increase throughput by allowing the system to Microsoft, Citrix, and Dell—to address these problems. concentrate on the data in the frames instead of the Broadcom controllers are included across a range frames around the data. However, because of the

68 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. single-threaded nature of the hypervisor in performance, especially at 10 Gbps; this VMware vSphere™ 4.1 platform on Dell processing I/O and the duplicate I/O copies problem is addressed by offloading these PowerEdge servers. in the virtualization layer, performance is tasks into a network adapter, where the Broadcom iSCSI HBA functionality still limited with this approach. Broadcom transport queue manager can transmit enables on-chip processing of the iSCSI also supports the VMware® NetQueue and packets from multiple queues and steer protocol (and of the TCP and IP protocols), Microsoft® Hyper-V™ VM queue (VMQ) the receive packets into multiple queues. helping free up host processor resources features, helping remove such single-queue The VMware NetQueue and Microsoft at 10 Gbps line rates over a single Ethernet bottlenecks and avoiding the need for Hyper-V VMQ features enable Broadcom port. This functionality provides extended stateful offloads such as TCP offload. And controllers to take on these tasks, thereby performance benefits that help meet the Internet SCSI (iSCSI) host bus adapter helping meet the demands of bandwidth- the demands of bandwidth-intensive (HBA) mode in Broadcom controllers, intensive applications that require high applications requiring high-performance supported as part of Dell virtualization levels of performance and networking block storage I/O for the hypervisor, solutions, can provide excellent throughput in virtualized environments. servicing all instances of the VM. performance in virtualized environments. Storage offload iSCSI boot Use of multiple queues Networked storage is crucial in virtualized iSCSI boot allows a server to boot an OS The trend toward increased processor environments, enabling the smooth over a storage area network (SAN), avoiding core density is leading to the use of an migration and failover of a VM from one the need for local disk storage—a primary increased number of VMs, which requires physical server to another. iSCSI has source of computer system failures. In additional processor cycles to route emerged as a high-performance, accessible addition to enhancing system reliability, the packets to the VMs. Using hardware networked storage technology popular in use of diskless servers helps simplify the IT queues provided by the network controller, many virtualization deployments. Broadcom administrator’s workload by centralizing the virtualization vendors have avoided the NetXtreme II® iSCSI HBA functionality, creation, distribution, and maintenance of single-thread limitation of a traditional OS with support for VMware, Microsoft server images; reducing the overall need and optimized the hypervisor for multiple Hyper-V, and Citrix® Xen® virtualization, for storage capacity through increased disk hardware threads. provides the converged functionality capacity utilization; and adding increased In both the VMware and Microsoft needed in virtualized environments by data redundancy through the use of data Hyper-V platforms, packets must traverse offering on-chip processing to help free mirroring and replication. the hypervisor or parent partition because up processor resources and increase As the use of SANs continues to grow in a direct path is not available between the bandwidth and performance, as shown virtualized environments, and as administrators controller and the VMs. On egress, packets in Figure 1. Broadcom GbE and 10GbE continue to recognize the advantages of are first copied from the originating VM iSCSI HBA functionality as shown in moving local storage from individual servers for processing in the virtual switch. The this figure is enabled by default in the to centrally managed storage arrays, network destination Media Access Control (MAC) address and virtual LAN (VLAN) ID are Signiﬁcantly reduced iSCSI processing in OS stack looked up to determine the route, and the beneﬁts the guest OS and virtualization platform packet is then copied to the receive queue VM 1 VM 2 VM n of the other VMs and/or submitted to the OS SCSI OS SCSI OS SCSI stack stack stack network driver for transmission. On ingress, packets are indicated to the switch, which VMware iSCSI stack Existing iSCSI Figure 1. Broadcom uses the destination MAC address and VLAN Hypervisor and iSCSI HBA mode Broadcom iSCSI or VMM in a virtualized ID to determine which VM or group of VMs network stack o oad driver environment the packets can be copied to. Route lookup, data copy, and filtering On-chip iSCSI tasks represent additional processor load o oad processing and latency absent in the non-virtualized Broadcom GbE/10GbE NIC Third-party GbE/10GbE NIC environment. The associated overhead On-chip iSCSI processing frees processor cycles, helping can significantly affect networking to reduce power consumption and increase performance

boot options such as iSCSI boot are poised to become having multiple virtual functions (VFs). A VF is a an increasingly common feature within the data center lightweight PCIe function, and resources associated and throughout the enterprise. Broadcom, VMware, with the main data movement of the function are Microsoft, and Citrix are working to create simple available to the VM. The VF can be serially shared yet richly featured iSCSI boot options that use iSCSI between different VMs—assigned to one VM, and to replace local storage in virtualized environments. then reset and assigned to another. In addition, a VF Broadcom GbE and 10GbE iSCSI boot functionality can be transformed from a VF into a PF. is enabled by default in VMware vSphere 4.1 on Dell Comprehensive support for PCIe SR-IOV PowerEdge servers, and Broadcom iSCSI HBA mode involves both enhancing existing capabilities and and iSCSI boot for 10GbE controllers in hypervisors are adding new capabilities to the platform and OS. both supported on Dell platforms.1 Network controller device drivers supporting SR-IOV also must be re-architected to support additional Phase 2: Integrating SR-IOV enhancements communication paths between PFs and VFs. The second phase of Broadcom’s enhanced virtualization effort, now under way, focuses on VMware VMDirectPath Single Root I/O Virtualization (SR-IOV); these High throughput and low latency are especially enhancements are expected to be available in important in a distributed system, wherein the nodes’ Dell systems using Broadcom NetXtreme II 57712 I/O latencies affect the performance of the cluster controllers. SR-IOV-capable Ethernet network and the overall application. Low latency is necessary controllers are designed to enhance I/O throughput to preserve data coherency in large database clusters and reduce processor utilization while increasing implementing scalable SR-IOV network adapters. scalability and devices’ sharing capabilities. SR-IOV Using the VMware VMDirectPath network plug-in allows the direct I/O assignment of an Ethernet architecture and an SR-IOV-capable Broadcom network controller to multiple VMs, helping maximize device, a VF can be directly assigned to a VM— the network adapter’s full bandwidth potential. yielding near-native performance, avoiding additional The PCI Express (PCIe) SR-IOV specification I/O copying in the hypervisor, and supporting a for PCI I/O virtualization and sharing technology is comprehensive array of virtualization features, the basis for the SR-IOV implementation in SR-IOV- including VMware vMotion™ technology and capable Broadcom controllers. The specification Microsoft Hyper-V live migration (see Figure 3). Direct defines an extension to the PCIe specification that assignment of PCI devices to VMs is necessary for I/O enables multiple system images or VMs to share PCIe appliances and high-performance VMs. hardware resources (see Figure 2). The Broadcom With dynamic VMDirectPath or Uniform Pass- SR-IOV device presents a physical function (PF) Through version 2 (UPTv2), the device interface is

VM 1 VM 2 VM n Virtual VF device VF device NIC driver driver

Figure 2. SR-IOV-capable Broadcom PF device Hypervisor controller in a virtualized environment driver or VMM VF VF PF

SR-IOV-capable Broadcom controller

1 For more information on iSCSI functionality in Broadcom adapters for Dell systems, see “A unified networking approach to iSCSI storage with Broadcom controllers,” by Dhiraj Sehgal, Abhijit Aswath, and Srinivas Thodati, in Dell Power Solutions, 2010 Issue 2, dell.com/downloads/global/power/ps2q10-20100411-broadcom.pdf.

70 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. VM 1 VM 2 VM n Virtual VF device VF device NIC driver driver

PF device Hypervisor Figure 3. VMware VMDirectPath with an driver or VMM SR-IOV-capable Broadcom controller VF VF PF

NIC embedded switch

SR-IOV-capable Broadcom controller

split in two, enabling pass-through of performance- Effective and efficient virtualization critical operations such as TX/RX producer index The network is a key component of virtualized registers, interrupt mask registers, and emulated environments, and network controllers that are infrequent operations in the management driver enhanced to meet the demands of virtualization running in VMware ESX. To implement live can help maximize performance. Broadcom’s migration, the VF is acquiesced and switched virtualization-focused efforts enable organizations to emulation mode from pass-through mode, using Dell systems to take advantage of offload allowing the minimal device state to be technologies and flexible, real-time I/O— checkpointed or restored. Most of the state lives in enhancements that facilitate effective and efficient the VM memory, and the guest operating systems virtualization. Broadcom capabilities in SR-IOV and are unaware of the migration. I/O pass-through functionality for Ethernet network Support for dynamic VMDirectPath requires controllers with TCP and iSCSI offload can provide re-architecting the OS platform and network device near-native performance and reduced latency. driver. The VMware platform implements a network Looking ahead, Broadcom plans to continue to plug-in architecture allowing pass-through of support and enable virtualization on Dell server performance-critical parts by partitioning the vmxnet platforms, helping IT organizations to accelerate driver to include a VM-specific shell and hardware- their adoption of virtualized infrastructures while specific module or network plug-in driver. The increasing performance and controlling costs. VM-specific shell implements the interface to the OS network stack and interacts with the hypervisor Dhiraj Sehgal is a senior product line manager for for configuration. The hardware-specific network Ethernet controllers at Broadcom. plug-in driver interacts with hardware in the data path and uses the VM shell interface for OS-specific Abhijit Aswath is a senior product line manager for Ethernet controller software at Broadcom. calls. VMware ESX controls the network plug-in used by the shell to load the plug-in into the VM based on Srinivas Thodati is a senior product marketing the VF and to map the VF into VM address space. manager for PowerEdge M-Series servers at Dell.

NIC embedded switch Learn more I/O virtualization and sharing are also required for point-to-point and switch-based configurations, Broadcom Ethernet controllers: enabling interoperability between VMs, VFs, chipsets, broadcom.com/products/Ethernet -Controllers switches, endpoints, and bridges. A Broadcom NIC embedded switch enables Ethernet switching between Dell PowerEdge servers: VMs, from VF to VF, and to or from external ports. dell.com/poweredge

Designing seamless Microsoft Exchange deployments across data centers

By Kong Yang, Jeff Sullivan, and Fred Johnson

In conjunction with key Microsoft® Exchange Server 2010 features, F5® BIG-IP® application delivery controllers can help organizations create high-performance, highly available messaging infrastructures spanning multiple data centers.

essaging applications have DAG brings up the passive database copy on become a critical part of enterprise another server and resumes services, helping operations. Microsoft Exchange minimize impact on end users. In Exchange Server M Server 2010 incorporates multiple 2010, DAG members can now be located at features designed to increase availability and geographically separated sites interconnected by a enhance management in these environments— wide area network (WAN), enabling organizations helping to maximize uptime while simplifying to use F5 BIG-IP WAN Optimization Module™ deployment, administration, and capacity sizing (WOM) technology to accelerate and encrypt DAG for IT administrators. replication traffic. Two key changes in Exchange Server 2010 F5 BIG-IP systems support a range of now enable organizations to incorporate F5 Exchange server roles, and can provide a variety BIG-IP application delivery controllers into of benefits depending on the specific deployment F5 Networks their Exchange infrastructures to help optimize and configuration (see the “Enhancing at the messaging traffic across multiple data centers. performance and availability” sidebar in this Dell TechCenter The first change is that end users no longer article). By designing an Exchange infrastructure The Dell TechCenter is connect directly to the Exchange Mailbox based on Dell™ and F5 BIG-IP systems, IT a collaborative online servers, even when using the Microsoft Office administrators can provide reliable e-mail and community that helps IT professionals solve problems, Outlook® e-mail client in native Messaging calendar access for end users while enhancing share best practices, and Application Programming Interface (MAPI) performance, availability, and management across learn. Visit the TechCenter Web site for much more on mode rather than through Remote Procedure their data centers. F5 BIG-IP systems—including Call (RPC) over HTTP. Instead, this access white papers, video demos, and more. is provided by the Client Access servers—an Availability and failover for Client Access approach that allows organizations to use F5 and Edge Transport servers delltechcenter.com/page/f5 BIG-IP systems to optimize and load balance F5 BIG-IP systems can help organizations end-user traffic. simplify and automate the management The second change relates to the Exchange of Exchange Server 2010 data recovery database availability group (DAG) capability, configurations across multiple sites. F5 BIG-IP which provides mailbox availability and automated Global Traffic Manager™ (GTM) systems provide database recovery in the event of failure. load balancing and failover across data centers, Continuous replication and monitoring of group automatically updating Domain Name System members enables quick detection and recovery (DNS) information based on application health. of a mailbox database; if an active member fails, Administrators can define policies that take

72 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. into account the real-time availability and WOM is designed to accelerate TCP traffic performance of Exchange Client Access and for data center applications and provide the Edge Transport servers, plan and easily initiate scalability to meet high-bandwidth requirements. local maintenance outages without disrupting In addition to symmetric compression and service, and maintain high availability even in the hardware encryption processing, WOM provides event of a disaster. symmetric deduplication, TCP optimizations, Effectively implementing failover between tunneling, and a single point of administration. data centers across a WAN may require Deduplication, a key acceleration feature, builds accelerating and encrypting DAG replication a cache on the device at each end of the WAN traffic: poor WAN conditions can cause data link, and when a duplicate pattern in the network transfer delays, and regulatory or compliance traffic is found, a small reference to the cache requirements often dictate the need for is transmitted instead of the entire pattern—an encryption. While GTM handles the data important way to help reduce the amount of center availability and failover through DNS, F5 WAN traffic. F5 TCP Express™ technology applies BIG-IP WAN optimization can accelerate and advanced protocol optimizations to the LAN and encrypt the DAG traffic, helping to secure the WAN to help deal with latency and packet loss data and shorten replication times between sites. Because DAG replication can therefore complete in less time than it would otherwise, the passive Mailbox databases are more likely to Enhancing performance be up-to-date when activated in the event of a failure of the primary database. and availability

F5 BIG-IP systems can offer a variety of advantages in Microsoft Exchange WAN optimization for DAG replication Server 2010 environments—helping administrators create an efficient, WAN conditions and remote data transfer high-performance, highly available infrastructure. performance can vary widely depending on factors such as available network bandwidth, Simplify site-to-site failover and shorten disaster recovery times link type and quality, and distance between sites. WANs that have limited bandwidth, high Minimize downtime in the event of a failure latency, and packet loss can cause delays for DAG Accelerate and encrypt database availability group (DAG) traffic over replication, and prolonged data transfer times between disaster recovery sites can put business wide area networks (WANs) continuity plans at risk. Scale out Client Access and Edge Transport servers Out of the box, Exchange Server 2010 provides compression and encryption for DAG Support consolidation to help simplify management and save on partners located on different IP subnets. These hardware use, rack space, and power consumption features are built-in, cost-efficient, and effective Enable server offload using BIG-IP systems to help reduce hardware at helping to reduce the amount of network requirements and increase virtual machine server density traffic and ensure the privacy of the transfers, but both are resource intensive and take away server Combine BIG-IP feature sets on a single platform, helping maximize processing power from the primary application. the value of hardware investments In addition, compression alone does not address poor performance associated with problematic Support multiple applications and multi-tenancy on a single BIG-IP WAN conditions. Specialized technologies such appliance as F5 BIG-IP WOM are required to effectively Protect Outlook Web App (OWA) services and support regulatory and overcome poor WAN replication performance Payment Card Industry Data Security Standard (PCI DSS) compliance and offload the tasks of compression and encryption from the Mailbox servers.

and increase the efficiency and reliability delivery technologies can provide end-to- Microsoft Cluster Service (MSCS)—running of network communications. Exchange end high availability for applications. NLB and MSCS on the same hardware is Server 2010 administrators needing to Administrators can load balance not recommended. implement DAG replication over WANs Client Access servers using Microsoft Scaling out an Exchange Server 2010 can benefit significantly from F5 WAN Windows® Network Load Balancing infrastructure requires the use of a hardware optimization technologies. (NLB) clustering, which provides a cost- load balancer for Client Access and Edge effective choice for some configurations. Transport servers. The advanced features Scaling and optimization for local Client However, administrators should consider and benefits of BIG-IP application delivery Access and Edge Transport servers using F5 BIG-IP Local Traffic Manager™ controllers extend beyond those of software Another critical aspect of a scalable, (LTM) application delivery controllers for load balancers to include security, advanced highly available infrastructure is environments that contain eight or more persistence methods, application health application delivery and load balancing Client Access servers, with servers running check monitors, high availability, Secure within a local data center. Working more than one Exchange server role, Sockets Layer (SSL) acceleration and together, global and local application or when Exchange servers are running offload, TCP optimization and offload, and

Simple Mail Transfer Protocol Remote users (SMTP) servers

IMAP4, POP3, HTTP, RPC over HTTP, MAPI over VPN Internet

Data center 1 Internal users Internal users Data center 2

F5 BIG-IP systems F5 BIG-IP systems Demilitarized DMZ zone (DMZ) Global Trac Manager Global Trac Manager Local Trac Manager Local Trac Manager

Edge Transport Edge Transport servers servers Microsoft Active Microsoft Active Directory® server Directory server

LAN LAN WAN

F5 BIG-IP systems F5 BIG-IP systems

Local Trac Manager WAN optimization Local Trac Manager Application Security Manager Application Security Manager WebAccelerator™ module WAN Optimization Module WebAccelerator WAN Optimization Module

Client Access servers Mailbox servers (DAG) Client Access servers Mailbox servers (DAG)

Dell PowerEdge™ servers

™ Dell EqualLogic storage Dell PowerEdge blade servers Dell EqualLogic storage

Dell Agile Consolidated model (1,000 mailboxes) Dell Agile Consolidated model (virtualized, 10,000 mailboxes)

Figure 1. Example Microsoft Exchange Server 2010 architecture based on Dell servers and storage and F5 BIG-IP application delivery controllers

74 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Web acceleration and offload with caching and compression—all of which enhance the end- user application experience and help reduce the Simplifying Exchange Exchange server loads. F5 BIG-IP LTM and GTM systems are designed to complement each other deployment to support high availability: LTM monitors the Client Dell Services can help organizations simplify and streamline Microsoft Access servers within a data center, while GTM Exchange Server 2010 deployments and ongoing management while monitors the health of the Client Access services reducing risk, cost, and implementation time. and the LTM systems across multiple data centers.

Together, they can make Exchange Server 2010 Dell consulting services and Exchange reference architectures help highly available both locally and globally. ensure that e-mail and calendaring platforms based on Exchange The Exchange Outlook Web App (OWA) Server 2010 are optimized, flexible, efficient, and secure. feature allows end-user access to Exchange messaging through a rich, full-featured interface Dell can provide a single point of contact for a comprehensive on major Web browsers. Because OWA is a e-mail and calendaring package based on Exchange Server 2010, Web service, it can also provide an attack target, including hardware, software, services, and support. opening critical messaging systems to risk of data loss, compromise, theft, denial of service, The result is an integrated, end-to-end messaging system designed and Web scraping, among other things. The for maximum productivity, efficiency, security, and scalability. F5 BIG-IP Application Security Manager™ (ASM) Web application firewall, which runs on BIG-IP systems alongside LTM and other modules, can “Simplifying Exchange deployment” sidebar in this help protect Exchange Web services from attack. article). Through this type of design, organizations Subscription-based updates are designed to can create integrated multi-data-center keep ASM current to help ensure regulatory and messaging environments while supporting site- Payment Card Industry Data Security Standard to-site failover, simplified disaster recovery, high (PCI DSS) compliance. ASM also helps simplify levels of security, and enhanced performance. audit efforts by producing an executive summary of requirements and recommendations for bringing Kong Yang is an evangelist for the Dell application environments into compliance. TechCenter online community.

High-performance, highly available Jeff Sullivan is a storage and Linux® OS evangelist for the Dell TechCenter online Exchange infrastructure community. Combining Dell servers and storage, F5 BIG-IP application delivery controllers, and Microsoft Fred Johnson is a strategic partner engineer with F5 Networks dedicated to Dell Labs. Exchange Server 2010 can help organizations create high-performance, highly available enterprise messaging infrastructures. Figure 1, Learn more for example, shows one possible design based F5 Networks and Dell: on Dell reference architectures for Exchange, f5.com/dell illustrating how these components can link two data centers across a WAN while taking advantage of F5 BIG-IP modules to help maximize Dell and Microsoft Exchange: dell.com/exchange performance. Dell Services can also work directly with organizations to identify appropriate architectures that can meet the specific needs F5 Networks and the Dell TechCenter on Twitter: of their environment while helping simplify twitter.com/f5networks deployment and ongoing management (see the twitter.com/delltechcenter

A smart approach to application intelligence

By Patrick Sweeney and Matthew Dieckman

The SonicWALL® E-Class Network Security Appliance (NSA) Series provides comprehensive, flexible control over Web-based applications and other software—helping organizations to maintain security without sacrificing productivity or performance.

eb-based applications have SonicWALL E-Class Network Security Appliance become increasingly important (NSA) Series can deliver comprehensive in enterprise environments, application intelligence and control in Dell™ Woffering the promise of enhanced hardware–based environments—enabling collaboration, increased productivity, and reduced organizations to gain the benefits of these costs by connecting people and exchanging applications while still minimizing risk. information in new and efficient ways. Yet many organizations are unaware of the risks that Security challenges in a Web app world come with these benefits. Social networking and In response to a rapidly evolving Web application– streaming media threaten to drain bandwidth and based environment, IT departments must control productivity while competing with mission-critical and secure Web 2.0, cloud-based SaaS, and SOA Dynamic security applications for resources and time. Although software while at the same time managing usage software as a service (SaaS) and service-oriented and blocking potential data leakage. In this video, Patrick architecture (SOA) enable new ways of conducting Today, it is common for employees to Sweeney outlines how SonicWALL network security business, they can also introduce new threats, just download files from Web 2.0 social networking appliances dynamically as personal Web-based e-mail, instant messaging sites such as Facebook; stream rich media from adapt as organizations and threats evolve. (IM), and FTP can create new conduits for leakage social media sites such as YouTube; transfer of sensitive data. information using personal e-mail accounts with www.sonicwall.com/secret Many IT departments may be unaware that Google Gmail, Yahoo! Mail, and other services; their current firewalls are incapable of scanning and download files using peer-to-peer (P2P) application-layer traffic; even firewalls just two or applications such as BitTorrent and Kazaa. Such three years old are typically unable to scan and uses of data networks can rob enterprises of defend against the full range of threats that exist bandwidth, productivity, and confidential data, today. Other organizations resort to restricting and could lead to regulatory noncompliance. employee use of Web applications altogether, The enterprise use of SaaS and SOA is also on abandoning the potential benefits these the rise, with an increasing number of enterprises applications offer. Either way, a next-generation considering the use of cloud computing in firewall can help administrators to regain control the near future. Cloud-based applications are of the application-centric network. often prime targets for continuous, persistent The ubiquity and complexity of Web criminal attack from sophisticated profit-driven applications require IT departments to take or politically motivated hackers. In addition, a smart approach to these challenges. The prioritizing bandwidth for mission-critical

76 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Non-business-related Business-related applications applications (Facebook, YouTube, (Microsoft SharePoint, BitTorrent, Kazaa, Skype) Oracle, SAP, Salesforce.com)

Enterprise network

HTTP Scanning, blocking, 2 1 and bandwidth HTTP over SSL (HTTPS) management SonicWALL application IM Internet signature SonicWALL TCP database E-Class E-mail NSA appliance Figure 1. SonicWALL next-generation firewalls provide administrators with flexible control over applications and other software 3 Blocked Bandwidth-managed non-business-related non-business-related applications applications (Facebook, (BitTorrent, Kazaa) YouTube, Skype)

1 High-performance engine scans for applications regardless of port, protocol, or HTTPS/SSL encryption 2 Database containing thousands of application signatures is continually updated

3 Application control through real-time application visualization oers streamlined management

cloud-based applications (such as Microsoft® to establish application identification, control, SharePoint®, Oracle®, SAP®, and Salesforce.com visualization, bandwidth management, data- software) is key to ensuring productivity. leakage prevention, and threat protection (see The increase in Web-based application Figure 1). With multiple layers of protection traffic has placed a heavy burden on network delivered at Gigabit Ethernet speeds, the NSA performance. Social networking traffic alone can E8500 helps identify and block threats without consume a significant amount of bandwidth, and compromising performance. streaming music and video traffic (such as video Application identification, control, and of major sporting events) can overwhelm mission- visualization enable policy-based blocking or critical application traffic. And non-work-related restriction of specific applications. Organizations applications such as social media and online may want to restrict access to applications such as games can drain worker productivity. bandwidth-intensive streaming audio, video, and The latest generation of Web applications P2P programs; restricted file types (such as .exe, takes advantage of communication techniques .pif, .scr, and .vbs); or unauthorized Web 2.0 sites, that share common ports, random port entries, Web browsers, or IM clients. For example, P2P and other evasive techniques along with Secure applications like BitTorrent can steal bandwidth, and Sockets Layer (SSL) encryption. These techniques certain downloaded files may contain malware. allow Web applications to evade network detection Often, P2P developers update their applications with and control by traditional firewalls, which are versions designed to evade firewall defenses. The limited to port- and protocol-based inspection. SonicWALL Application Intelligence and Control feature set can restrict these dynamically changing Comprehensive application intelligence P2P applications through automatic updates of the Overcoming these multiple challenges can be application signatures. daunting for IT administrators. To help address The Application Intelligence and Control their concerns, SonicWALL has engineered capabilities can also secure bandwidth priority next-generation firewall appliances—such as the for critical applications and help control data E-Class NSA E8500—that enable administrators leakage. For example, administrators could define

policies to prioritize bandwidth availability by date (for example, policy settings, session management, and comprehensive audit prioritizing sales-related applications at the end of each quarter), or logs. In addition, the SonicWALL Global Management System to detect, block, or limit the bandwidth available to streaming video (GMS) enables centralized management of multiple SonicWALL and audio sites and file downloads; they could also allow members security appliances across multiple sites. of a predefined Microsoft Active Directory® group for marketing • Enhanced security: SonicWALL appliances help defend against to have access to Twitter or YouTube for promotional activities sophisticated, dynamically evolving Web-based threats, including while still restricting others’ access. To help control data leakage, P2P programs, e-mail attachments, and Web 2.0 executable administrators can define policies to block unauthorized outbound files. SonicWALL Application Intelligence is continually updated, transmission of sensitive, proprietary, or watermarked data over FTP leveraging more than 2,800 unique application signatures. In uploads, as attachments to enterprise Simple Mail Transfer Protocol addition, the SonicWALL Comprehensive Gateway Security (SMTP) and Post Office Protocol 3 (POP3) e-mail, and even through Suite (CGSS) offers not only Application Intelligence capabilities, personal Web-based e-mail. but also the SonicWALL Gateway Anti-Virus, Anti-Spyware, Equally significant, SonicWALL anti-malware and content and Intrusion Prevention Service (IPS) features; the SonicWALL filtering features can help protect against new and evolving channels Content Filtering Service feature; and more. for emerging threats. For example, a compromised Facebook page can suggest that a “friend” click a link to launch what appears to be Powerful, versatile protection a YouTube video—but is actually malware. SonicWALL’s patented Web-based applications are now leveraged by many enterprises— Reassembly-Free Deep Packet Inspection™ (RFDPI) engine can detect but in addition to expanding possibilities for productivity, these that the file is malware and prevent it from downloading, thereby applications also introduce specific security and productivity helping to protect both end users and the enterprise network. challenges. SonicWALL Application Intelligence extends the protection of SonicWALL network security appliances beyond Seamless integration blocking traditional network threats to the management and control Unlike “application-control only” appliances, SonicWALL appliances of data and applications that pass through the appliance. Application seamlessly integrate application intelligence with intrusion Intelligence capabilities are included in the NSA E8500 appliance prevention and advanced firewall defenses, forming a unified and and are available as a part of the CGSS suite, which includes the comprehensive network security tool that helps ensure threat SonicWALL Gateway Anti-Virus, Anti-Spyware, IPS, and Application protection, regulatory compliance, network performance, employee Intelligence features on SonicWALL NSA Series, E-Class NSA Series, productivity, and cost-effective management. Key advantages and TZ 210 appliances. By employing SonicWALL E-Class NSA include the following: security appliances such as the NSA E8500 in Dell hardware–based environments, IT departments can focus on ways to control these • Regulatory compliance: Regulations can make Web-based applications while still empowering end users to make effective use application attacks particularly onerous for financial, health care, of them—all without compromising performance. and application service providers, as well as for e-commerce businesses. SonicWALL Application Intelligence utilizes RFDPI and Patrick Sweeney is vice president of the product management team a dynamically updated signature database to help protect Web at SonicWALL, where he oversees its network security, content security, applications, including SSL virtual private network (VPN) portals. business continuity, and policy and management product lines. • Network performance: Network performance is crucial for Matthew Dieckman is the product manager for Secure Remote maintaining productivity, quality of service, and service-level Access solutions at SonicWALL, where he drives the development agreements. SonicWALL Application Intelligence can allocate road map for SonicWALL SSL VPN products and services. dedicated throughput levels for designated mission-critical applications, end-user groups, or times of day. SonicWALL

appliances with this capability feature connection multiplexing, Learn more which enhances the performance of protected Web sites and helps significantly reduce transactional costs. SonicWALL E-Class NSA E8500: • Streamlined management: Simplified management helps www.sonicwall.com/8500 to lower overhead and increase productivity. SonicWALL appliances ease management with a robust, intuitive Web-based SonicWALL Application Intelligence: management interface, automatic signature updates, flexible www.sonicwall.com/applicationintelligence

raphics processing units (GPUs) Expanding the were originally designed to make the massive calculations required for G rendering 3D images to a display. boundaries of Because of the nature of processing and creating images today, GPUs must have a large number GPU computing of cores that work in parallel to render models in photo-realistic detail. The growth of the gaming market, both for PCs and for gaming consoles, has Supporting up to 16 PCI Express devices in a flexible, driven a rapid pace of technological improvement, while the commodity nature of the gaming market highly efficient design, the Dell™ PowerEdge™ C410x has helped reduce the price of GPUs. expansion chassis helps organizations take advantage Researchers quickly discovered that GPUs could of the next step in high-performance computing also be exploited for high-performance computing (HPC) applications to deliver potentially massive architectures: GPU computing. increases in performance. They found that in HPC application areas such as life sciences, oil and gas, and finance, GPUs could dramatically increase the computational speed of modeling, simulation, imaging, signal processing, and other applications— with some seeing software running up to 25 times faster than on conventional solutions. “GPUs fundamentally offer much higher performance in servers,” notes Sumit Gupta, senior manager of Tesla™ products for NVIDIA, the company that invented the GPU and a leader in GPU computing. “And they offer this higher performance with lower overall power usage. This makes the performance per watt or per transaction of GPUs very compelling to IT managers deploying data center systems.” Gupta notes that on the Linpack benchmark, which is used to judge performance for the TOP500 Supercomputing Sites list, systems that use both GPUs and CPUs typically outperform systems based solely on CPUs. “The performance on Linpack was eight times higher using a server with two GPUs and two CPUs compared to the same server with just Flexible power two CPUs,” he says.1 “And in real applications, we’ve The Dell PowerEdge C410x provides a high-density seen the GPU deliver even greater performance chassis that connects 1–8 hosts to 1–16 GPUs advantages over servers with only CPUs.” and incorporates optimized power, cooling, and The ability to deliver dramatic increases in systems management features. compute performance at a reduced cost has positioned GPU computing at the forefront of the • Up to 16.5 TFLOPS of computing throughput next wave of HPC architecture adoption (see • Hot-pluggable components for simplified serviceability 1 Based on NVIDIA testing using the Linpack benchmark to compare a 1U server • Highly efficient design to help minimize energy with two quad-core Intel® Xeon® X5550 processors at 2.66 GHz and 48 GB of RAM against the same server with two NVIDIA Tesla M2050 GPUs, two use and costs quad-core Intel Xeon X5550 processors at 2.66 GHz, and 48 GB of RAM.

of its first GPU solutions with the National Center for Supercomputing Applications on Dell Comparing CPUs with GPUs PowerEdge servers through hardware interface cards in PCI Express (PCIe) slots, validated with Central processing units (CPUs) are highly versatile processors with large, NVIDIA® interface cards (see the “Maximizing complex cores capable of executing all routines in an application. They are supercomputer performance and efficiency” used in the majority of servers and desktop systems. Compared with CPUs, sidebar in this article). graphics processing units (GPUs) are more focused processors with smaller, Now, Dell is helping make GPU processing simpler cores and limited support for I/O devices. power even more accessible through the Dell Recent generations of GPUs have specialized in the execution of the PowerEdge C410x PCIe expansion chassis, which compute-intensive portions of applications. They are particularly well suited enables organizations to connect servers through for applications with large data sets. Application development environments the appropriate host interface card to up to 16 external GPU cards. On the measure of peak for GPUs use techniques that allow the GPU to handle compute-intensive single-precision floating point performance, a portions of applications that usually run on CPUs. PowerEdge C410x with 16 NVIDIA Tesla M2050 GPU modules can deliver up to 16.5 TFLOPS of computing throughput. the “Comparing CPUs with GPUs” sidebar in this The impetus for the creation of the article). GPU programming methods and toolkits PowerEdge C410x came from an oil and gas have advanced, making it easier than ever before company that wanted to accelerate processing for software developers to take advantage of GPU speeds for the complex seismic calculations computing. And in complex and computationally used in the search for oil reservoirs, notes Joe intense environments, GPU performance and Sekel, a systems architect with the Dell Data speed of delivery can contribute to important Center Solutions (DCS) team.“Given the industry outcomes—such as finding a cure in biomedical they are in, they are focused on getting to their research or modeling and predicting the path and answers as fast as they can,” he says. “They are intensity of the next hurricane. very motivated to use all means to accelerate the answer.” Accelerating processing speeds In particular, the company wanted to Dell has been working toward accessible GPU investigate its options for increasing the ratio of

The Dell PowerEdge C410x computing for several years. Dell provided GPUs to CPU sockets in its x86-based servers was designed from the ground technology for high-performance GPU to help speed application throughput. “This up to efficiently power and cool ™ 1–16 PCIe devices, with the flexibility clusters with the Dell Precision R5400 rack company was currently running with two GPUs of connecting to 1–8 hosts workstation, and in 2008, Dell delivered some per two-socket server,” Sekel says. “However, they were projecting that if they kept tweaking their code, they could potentially bump that ratio up to four GPUs per two-socket server, so they could get to the answer faster. But that wasn’t something they were ready to do quite yet.” The problem was that the company wasn’t sure of the right ratio. That’s because its ability to use the additional GPU processing power in its x86-based servers depended to a large degree on the ongoing optimization of its algorithms and software. So it didn’t want to lock itself into a specific configuration. In response, Dell DCS system architects set off on a path that ultimately led to the

80 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. development of the PowerEdge C410x. In addition to offering the flexibility to change the number of GPUs over time and to share GPUs among multiple host servers, the chassis also addresses fundamental problems that HPC users encounter when they add PCIe devices to existing servers. In simple terms, today’s dense, power-efficient servers have a limited ability to accommodate additional PCIe devices. “Today’s servers are very optimized around density for x86 computing,” Sekel says. “Everything we do in there in terms of packaging, power, and the fan subsystem is really honed for maximum density given that particular set of components. We didn’t want to compromise server density by putting GPUs in the chassis. So this pointed to the need for an expansion chassis that talks to servers over PCIe.” Moving PCIe devices out of servers allows them to maintain density, power, and thermal supplies are individually serviceable while the Hot-add PCIe modules along with efficiency without sacrificing performance, while chassis is in use—meaning that IT staff can pull hot-plug fans and power supplies in the Dell PowerEdge C410x the purpose-built external expansion chassis helps individual components from the chassis for make it easy to service individual optimize power and cooling for PCIe devices servicing without taking the entire unit down. components such as GPUs. In addition, the use of an external “Given that the chassis and the GPUs in it are PCIe expansion chassis provides the flexibility shared by multiple hosts,” Sekel says, “the last to accommodate a wide variety and increased thing you want to have to do is take down quantity of PCIe devices used with servers. the entire chassis when you need to service a single component.” Designing for a wide range of applications While delivering this high level of The PowerEdge C410x is a 3U external PCIe serviceability, the PowerEdge C410x also helps expansion chassis that allows host server nodes reduce costs. These savings stem from the to connect to up to 16 PCIe devices; each increased density, the reduced weight of the individual host server can access up to 4 PCIe chassis, and the reduced requirements for devices in the chassis. Although the chassis switches, racks, and power compared with has optimized power, cooling, and systems competitive configurations. management features, it does not have CPUs Sekel considers the PowerEdge C410x or memory. It simply provides optimized power to be well suited for a wide range of HPC and cooling in a shared infrastructure to support applications, including oil and gas exploration, GPUs and other PCIe-based devices such as biomedical research, and work that involves solid-state drives, Fibre Channel cards, and complex simulations, visualization, and InfiniBand cards. The chassis also supports mapping. The PowerEdge C410x is also a good redundant power supplies and redundant fans. choice for companies that work in gaming or “Aside from the flexibility, and the fact in film and video rendering, as well as those that we’ve put the GPUs in a high-density box that simply require additional PCIe slots in an that’s optimized for power and cooling of existing server, Sekel says. GPUs, we provide a serviceability model that The chassis is currently offered with the is fairly unique in this space,” Sekel notes. The NVIDIA Tesla M1060 and M2050 GPU modules, hot-pluggable PCIe modules, fans, and power with the Tesla M2070 expected to be added in

Maximizing supercomputer performance and efficiency

The National Center for Supercomputing Applications requirements. “The compute power density is a lot higher with (NCSA) is at the forefront of GPU computing. One of its the GPUs,” he says. “They also have much greater heat density. supercomputers, named Lincoln, is a 47 TFLOPS peak The advantage is a smaller footprint and an attained performance computing cluster based on Dell hardware with NVIDIA per watt that is much greater than that of traditional CPUs. While GPU units and conventional Intel CPUs. By mixing GPUs and there are some challenges in being able to cool and provide CPUs, Lincoln broke new ground in the use of heterogeneous power, GPUs are more cost-effective because the total power processors for scientific calculations. per flop and total cooling per flop are less.” This combination allows NCSA to take advantage of Towns offers the example of the Amber molecular the cost economies and extreme performance potential of dynamics application, which an estimated 60,000 academic general-purpose GPUs (GPGPUs), notes John Towns, director researchers use for biomolecular simulations. “For that of persistent infrastructure at NCSA. “What we’re seeing, for application, researchers are realizing in the neighborhood of the applications that have emerged on GPUs, are applications 5 to 6 gigaflops per watt. The thing to keep in mind is that that on a per-GPU basis have an equivalent performance of most of the time when you’re talking about this with respect anywhere from 30 to 40 CPU cores all the way up to over for CPUs, you’re talking about megaflops per watt. And that’s 200 CPU cores,” he says. “So this makes GPU platforms realized performance, not peak performance. So the realized anywhere from 5 to 50 or more times more cost-effective performance for this application on CPU cores is more on the than a CPU-only-based computing platform.” order of 300 to 400 megaflops per watt, as opposed to 10 Towns also notes that GPU-based systems have distinct to 20 times that on a GPU. So it makes a big difference when advantages over CPU-based systems in terms of total cost of it comes to considering total cost of ownership in delivering ownership, stemming from their reduced power and cooling resources to a broad research community.”

fall 2010. These Tesla 20-series modules are based many data centers. ECC corrects errors that can on the next-generation Compute Unified Device happen on the memory, and CPUs have had ECC Architecture (CUDA) GPU architecture (code-named for many years now. So adding ECC to GPUs is a “Fermi”), and are designed to support the integration very big thing.” of GPU computing with host systems for HPC and In another important advance, the Tesla large, scale-out data center deployments. 20-series modules offer Level 1 and Level 2 cache Compared with previous-generation NVIDIA memory, which helps increase system performance GPUs, the Tesla 20-series modules offer higher by reducing latency, Gupta says. These two levels throughput, according to Gupta. On the measure of cache also give programmers increased flexibility of double-precision floating point performance, in how they write programs for GPUs. the Tesla 20-series modules are rated to deliver The PowerEdge C410x is qualified with the more than 10 times the throughput of a quad-core PowerEdge C6100 server, but is designed to connect x86-based CPU. The Tesla 20-series modules also to any server with the appropriate host interface card. offer the benefits of error-correcting code (ECC) In addition, although it initially targets the NVIDIA memory for increased accuracy and scalability, Tesla M1060 and M2050 GPU modules, the chassis Gupta says. can accommodate a variety of PCIe-based devices “This is the first time anyone in the industry beyond GPUs, including network cards and storage has put ECC on a GPU,” he notes. “These are data devices—so the options for the chassis are expected center products, and ECC is a requirement in to grow significantly over time.

How does a

Supporting GPU development The GPU industry as a whole is working actively to support GLoBaL consumer the efforts of organizations moving toward GPU computing. Software developers who want to create code for GPUs packaGed Goods can take advantage of an ever-widening range of resources, including off-the-shelf compilers, tools, and libraries for GPU Leader proTecT programming, along with hundreds of available applications. NVIDIA, for example, provides compilers and libraries for its CUDA parallel computing architecture, which supports oVer 300 Brand standard application programming interfaces such OpenCL and Microsoft® DirectCompute as well as high-level programming secreTs? languages such as C/C++, Fortran, Java, Python, and the Microsoft .NET Framework. NVIDIA also maintains an online resource site, CUDA Zone, for GPU developers; programmers can visit the site at nvidia.com/cuda to obtain drivers, a CUDA software development kit, and detailed technical information. The academic community is also moving into GPU computing, Gupta notes; more than 350 universities now offer courses in GPU computing. Looking ahead, Gupta sees GPUs playing an increasingly prominent role in computing as developers learn to take advantage of this parallel processing technology. “This will be for both classical scientific computing tasks and enterprise needs,” he says. “Today, the major use of GPUs is in scientific computing. But we are starting to see GPUs become more relevant to the traditional enterprise data center—for business analytics, for example. Business analytics tasks run very well on the GPU.”

Enabling accessible GPU computing In HPC environments, GPU computing offers one of today’s most powerful computational technologies on a price/performance basis. To help organizations extend their use of GPU computing, Dell offers IT consulting services, rack integration (United States only), on-site deployment, and support services for organizations deploying and using Web 2.0, virtualization, cloud, mobility—market GPU-based Dell systems. Taking advantage of these services leaders continuously adopt new technologies, and systems like the PowerEdge C410x expansion chassis can exposing organizations and their “secret sauce” help organizations dramatically increase performance while to new threats. SonicWALL’s network security is intelligent enough to adapt as organizations and maximizing efficiency. threats evolve—dynamically, globally. Learn more at sonicwall.com/secret

Learn more

Dell PowerEdge C-Series: dell.com/poweredgec

NVIDIA Tesla GPUs: nvidia.com/tesla DYNAMIC SECURITY FOR THE GLOBAL NETWORK

Customer Advocacy Ad_Dell Power Solution Magazine 4.025"x 10.875", +0.125" bleed on 3 sides SW1024 High-performance computing

Optimizing HPC clusters with 10 Gigabit Ethernet iWARP technology

By Tom Stachura

As high-performance computing experts push the limits of cluster efficiency and price/performance, Dell™ servers in an iWARP-enabled Intel® 10 Gigabit Ethernet fabric can help reach new performance and efficiency levels—and a TOP500 ranking.

ew supercomputers are continually challenges. Now, Internet Wide Area RDMA Protocol coming online as business and academic (iWARP) provides a way to deliver Remote Direct researchers strive to keep up with Memory Access (RDMA) clustering on 10 Gigabit Never-growing scientific and engineering Ethernet (10GbE) network adapters—an approach that computational demands. In the persistent quest to do enables cluster users to overcome fabric challenges more work with greater efficiency and less cost than and continue to push the supercomputing envelope. ever before, high-performance computing (HPC) has moved from mainframe computers to economical and Assessing HPC connectivity challenges easily maintained clusters, including clusters made up of InfiniBand connectivity presents several challenges cost-effective servers based on the x86 architecture. for supercomputing. First, InfiniBand is a highly Today, some of the world’s most powerful specialized switched-fabric communications link supercomputers are composed of x86-based clusters that requires special expertise and tools for setup, with performance delivered on the network side configuration, and management, and costs for tools, using high-speed InfiniBand connectivity. But for training, and outside expertise can be high. Second, HPC experts striving toward reduced fabric costs because Ethernet is the de facto standard fabric for and simplified use, InfiniBand can present several local networking traffic, using InfiniBand requires

84 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. the HPC team to maintain two networking sources collectively account for nearly 100 percent technologies—InfiniBand connectivity to run of processor overhead related to networking. the application traffic and Ethernet connectivity The iWARP extensions help reduce processor Understanding to manage the servers in the cluster. Third, overhead, memory bandwidth utilization, and iWARP InfiniBand remains relatively costly in terms of latency using several advanced techniques: In this Intel technology per-port pricing, so hardware costs can be an brief, learn more about how impediment to increasing cluster cost-efficiency. • Kernel bypass (OS bypass): Context switching iWARP helps reduce Ethernet Despite these factors, InfiniBand connectivity has can be a costly process in terms of overhead. overhead and latency to enable organizations to gain been a logical technology for HPC because of its By allowing the application to communicate the full network throughput high throughput, low latency, and scalability. directly with the network controller, iWARP benefits of 10 Gigabit Ethernet.

Ethernet, in contrast, is widely used for bypasses the kernel. This permits the kernel-to- download.intel.com/support/ connecting users and network resources, but user context switches to be avoided and helps network/sb/understanding _iwarp_final.pdf has not been traditionally preferred for low- reduce latency and processor load. latency supercomputing. Ethernet does have • Intermediate buffer copies avoidance: Data the advantage of being extremely cost-effective is placed directly in application buffers rather for general-purpose LAN traffic. The bandwidth than being copied multiple times to driver and of Ethernet has increased 10-fold with the network stack buffers, which helps reduce mainstream availability of 10GbE networking latency as well as memory and processor use. hardware, providing viable connectivity • Transport (TCP/IP) processing acceleration: performance for HPC clusters. And Ethernet TCP/IP processing is performed in hardware has broad scalability, providing the ability to instead of the OS network stack software, dynamically add and remove nodes in cluster enabling reliable connection processing at environments. speed and scale. Data management and However, achieving low latency and high network protocol processing can be executed bandwidth for HPC applications requires on the Ethernet adapter, which provides enhancements to standard Ethernet. Standard hardware acceleration of the process. Ethernet communicates using a kernel network protocol stack that adds overhead in terms of compute load, memory bandwidth, and network Without iWARP With iWARP latency. In HPC environments, this overhead can greatly reduce performance, and is therefore Application unacceptable.

I/O library Avoiding Ethernet overhead barriers User with iWARP Kernel A full implementation of currently available iWARP OS stack technology helps avoid virtually all processor Avoids copies networking overhead, returning those cycles to System driver and kernel the application. The Internet Engineering Task Software Force (IETF) standardized the iWARP specification Hardware in 2007; this standard specifies a set of extensions PCI Express (PCIe) to the TCP/IP protocol that define a transport Basic iWARP mechanism for RDMA. As such, iWARP provides a network iWARP network controller controller low-latency means of passing RDMA over Ethernet ﬂow TCP/IP ﬂow (see Figure 1). Together, these extensions address Ethernet the three major sources of networking overhead: application context switches, intermediate buffer copies, and transport (TCP/IP) processing. These Figure 1. Network controller data flow with and without iWARP

In fact, a large biomedical research facility has achieved excellent performance and near-linear TOP500 Supercomputing Sites scalability using iWARP-enabled 10GbE adapters on a rankings for iWARP cluster of 4,032 cores, as measured using the High- Performance Linpack (HPL) benchmark.1 The cluster The TOP500 Supercomputing Sites list ranks the most powerful supports large-scale workloads in a range of critical high-performance computing (HPC) clusters in the world using the High- research areas, including bioinformatics, image Performance Linpack (HPL) benchmark for distributed-memory computers— analysis, and sequencing, and was ranked at number helping provide a reliable basis for tracking and detecting trends in HPC. The list is 208 on the June 2010 TOP500 Supercomputing compiled by HPC experts at the University of Mannheim in Germany, the University Sites list (see the “TOP500 Supercomputing Sites of Tennessee at Knoxville, and the National Energy Research Scientific Computing rankings for iWARP” sidebar in this article). Center (NERSC) at Lawrence Berkeley National Laboratory. Rankings are updated The cluster consists of 14 server racks with twice a year, coinciding with the annual Supercomputing Conference (SC) in 36 servers per rack, for a total of 504 servers. The November and the International Supercomputing Conference in June. compute nodes are two-way Dell PowerEdge™ Several clusters on the list use Dell servers, including one 4,032-core R610 servers based on the Intel Xeon® processor cluster at a large biomedical research facility that uses Internet Wide Area X5550 architecture at 2.66 GHz, with 24 GB RDMA Protocol (iWARP) and 10 Gigabit Ethernet (10GbE) technologies. The of RAM and one 80 GB Serial ATA (SATA) hard June 2010 list ranked this cluster at number 208 for performance, achieving drive in each server. Cluster RDMA network higher performance than many of the world’s top clusters. Reranking for connectivity is provided by Intel’s NetEffect efficiency shows that this cluster comes in at number 84, boasting a TOP100 iWARP-enabled 10GbE server cluster adapters. efficiency—the highest of all the listed Ethernet-based supercomputers. For more At the rack level, each server has two information on the TOP500 list, visit top500.org. connections to one of two 48-port Arista 7148SX switches: one 10GbE link (using direct-attach twinaxial cable) for RDMA traffic and one Gigabit Using iWARP-enabled 10GbE fabrics Ethernet (GbE) link for all other traffic. Each switch in HPC clusters has eight 10GbE uplinks (16 per rack) to a group of Using these techniques, the iWARP standard additional switches. Software running on the cluster enables low-latency network connectivity that can includes the Red Hat® Enterprise Linux® 5.3 OS, be well suited for HPC clusters. Intel’s NetEffect™ OpenFabrics Enterprise Distribution (OFED) 1.4.1, iWARP-enabled 10GbE server cluster adapters are and Intel Message Passing Interface (MPI) 3.2.1. specifically designed to deliver this capability in In October 2009, project engineers running HPC environments. this cluster in the lab with the HPL benchmark

40,000 100

90 35,000 80 30,000 70 Figure 2. Performance and 25,000 60 efficiency test results for an HPC cluster using an iWARP-enabled 20,000 50 10GbE fabric 40 15,000 30

10,000 E ciency (percent) Performance Performance (GFLOPS) Performance 20 5,000 E ciency 10

0 0 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 Number of cores

1 For more information on HPL, visit netlib.org/benchmark/hpl.

86 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. attained performance of up to 35.81 TFLOPS management tools, and solution stacks. Using at 84.14 percent efficiency (see Figure 2). An mainstream Ethernet connectivity for compute HPL problem size of 1,200,000 was used, and clusters can now provide highly favorable the problem size necessary to achieve half the performance, efficiency, and scalability. Taking Tom Stachura is the product performance (n/2 problem size) was 300,000. The advantage of iWARP-enabled 10GbE adapters line manager for HPC Ethernet performance data scales in a nearly linear fashion allows RDMA traffic to be passed effectively over an products at Intel. He has over as the number of cores applied to the benchmark Ethernet fabric. Organizations can now obtain the 17 years of industry experience workload increases. From an engineering low latency that HPC clusters need while capitalizing in engineering, architecture, strategic planning, and product perspective, the linearity of scaling in the results on the ease of use and familiarity of Ethernet. marketing. helps ensure the viability of the topology for large- Using Ethernet connectivity as a unified scale computational problems. fabric for cluster interconnects, LANs, and Based on these results, the cluster ranks at storage can help lower total cost of ownership Learn more number 84 for efficiency compared with other by significantly reducing the number of switches TOP500 clusters—a level more efficient than and cables required. As 10GbE products and Intel Ethernet server adapters: many InfiniBand clusters, and the highest of the technology—including switches with high intel.com/go/ethernet listed Ethernet solutions. port density and technologies to further drive intelethernet-dell.com down latency—continue to evolve, future Maximizing the advantages of iWARP work can provide additional value in building Dell PowerEdge and Ethernet supercomputing platforms on Dell PowerEdge servers: dell.com/poweredge A key advantage of iWARP-enabled networking is servers interconnected with iWARP-enabled Intel its compatibility with existing network infrastructure, 10GbE adapters.

GEt thE MoSt

nEw Intel® Ethernet BanDwIDth out of 10 Gigabit X520 -x/k Your Data CEntEr. Dual Port Mezzanine Card The new Intel® Ethernet server and blade adapters feature advanced I/O virtualization, native iSCSI acceleration, and iSCSI Boot technologies for customers who need high levels of performance for their LAN and storage connections. Intel® Virtualization Technology for Connectivity reduces I/O bottlenecks in a virtualized server and improves performance, flexibility, and virtual machine scalability. Optimized for virtualization, Intel’s native iSCSI acceleration solution delivers iSCSI performance using customer-preferred, native iSCSI initiators and intelligent stateless offloads, making it the easiest and most cost-effective way to connect servers to iSCSI SANs. Intel® 10GbE and multi-port GbE server adapters on Dell™ PowerEdge™ servers—visit www.IntelEthernet-DellPS.com to learn more.

Intel® Ethernet Intel® Ethernet Intel® 10 Gigabit Xf Sr Intel® Gigabit Et Intel® Gigabit Et Intel® Gigabit Et 10 Gigabit X520-Da2 10 Gigabit at Server adapter Dual Port Server adapter Quad Port Server adapter Quad Port Mezzanine Card Dual Port Server adapter Server adapter for M-Series Blade Servers ©2010, Intel Corporation. All rights reserved. Intel and the Intel logo are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries.

50110126-intel.indd 1 8/30/10 2:21 PM Systems management

Managing the layers of a virtualized environment

By Ed Casmer

In virtualized environments, effective systems management encompasses much more than just virtual machines. Altiris™ Server Management Suite from Symantec enables a comprehensive approach to help streamline and simplify day-to-day operations.

lthough virtualization promises the VMs running on? A hypervisor—a pseudo-OS significant benefits, it often requires that brings its own set of concerns and capabilities, IT administrators to reexamine their distinct from those of a traditional Microsoft® A approach to systems management. Windows®, Linux®, or UNIX® OS. And what is the How does the virtualization model affect typical hypervisor running on? Hardware—yet another best practices? How should administrators layer in the virtualized stack. Each of these three approach patching, inventorying, and deploying layers requires ongoing management to help Altiris Server systems in a virtual world? Although it may seem ensure optimal performance (see Figure 1). Management that virtualizing an entire infrastructure would Properly running hardware serves as the Suite in action eliminate many management concerns, experience foundation for a high-performance virtualized has shown that the same problems still exist—they infrastructure. Firmware updates, BIOS updates Visit Symantec online to view detailed video simply shift from one category to another. and settings, driver updates, and physical device demonstrations and to learn The Symantec™ Management Platform is settings are typical hardware-level concerns, more about how Altiris Server Management Suite designed with these issues in mind: instead and some management systems may be can enhance virtualization of building a new set of virtualization-specific unable to address this often vendor-specific management. tools, Symantec has elected to extend existing aspect of managing infrastructure. It may also dell.symantec.com/psq tools such as Altiris Server Management Suite to be difficult to identify needed updates and adapt and specialize their behavior when they configuration options at this level—requiring encounter virtual resources, allowing continued someone to manually identify and download standardization on existing tools and processes. each update if the management tools cannot By abstracting away the differences between automate this process. The Dell™ Management virtual and physical resources and supporting Console plug-in for the Symantec Management automated policies governing such differentiation, Platform allows administrators to address these these tools can help reduce the overall time hardware-level concerns, helping them keep required for routing management and free Dell PowerEdge™ servers up-to-date while administrators to focus on other initiatives. managing other physical components such as Dell Remote Access Controllers (DRACs) or Understanding the layers baseboard management controllers (BMCs). Managing a virtualized environment requires first The hypervisor is the next layer of the understanding its components. It’s tempting to virtualized environment, and comes with its think of the virtual machines (VMs) as constituting own deployment, patching, configuration, and the whole virtualized environment, but what are monitoring concerns. Managing the hypervisor

88 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. VM VM Agents Agents Agents Agents Granular control VM layer Apps Apps Apps Apps Figure 1. Effective systems OS OS management in virtualized environments encompasses the hardware, hypervisor, and virtual Hypervisor VMware Microsoft Linux Physical or machine layers layer ESX Windows or UNIX virtual platforms

Bare metal Hardware Physical server (includes BIOS, DRAC, BMC, and RAID conﬁguration)

breaks down into two main tasks: deploying it in Altiris Server Management Suite can help a consistent, standardized fashion, and keeping it address these questions. For example, Altiris maintained and up-to-date. Just as with hardware, Deployment Solution, included with Altiris Server keeping the hypervisor updated generally Management Suite, can manage automated offers distinct advantages, including bug fixes, deployments that include BIOS, DRAC, and performance increases, or entirely new features. RAID configuration; hypervisor deployment; VM Altiris Server Management Suite integrates with creation; OS deployment; and the installation of Dell Management Console and adds support for applications and needed agents, all triggered by managing not only the hypervisor, but also the a single drag-and-drop operation. Altiris Server next layer in the stack—the VMs. Management Suite also includes integrated

Once the VMs are up and running, however, workflow tools allowing the implementation of Ed Casmer is a technical does that mean that the infrastructure is a a server deployment process—complete with strategist in the Partner Alliance Group at success? VM templates may have simplified e-mail approvals and procurement authorizations Symantec dedicated to deployment tasks, but administrators must also to help protect against VM sprawl—as well as a the Dell Alliance. consider how they will patch the VM operating VM manager dashboard providing virtualization- systems, inventory installed software, and monitor specific views into both Microsoft and VMware® performance and configuration drift over time. In virtual environments from a single location. Learn more fact, most of the concerns attached to managing Altiris Server a physical system throughout its life cycle also Enabling a comprehensive approach Management Suite: apply to the VMs. to systems management symantec.com/business/ server-management-suite IT staff spend much of their time on routine Simplifying and centralizing management operations, and implementing a streamlined, Dell and Symantec: Understanding these three layers, as well as their comprehensive approach to systems dell.com/symantec specific management needs, is the key to gaining management can enable them to focus on symantec.com/dell and keeping control over the environment. At strategic initiatives to help their organizations minimum, administrators should consider several succeed. Altiris Server Management Suite from key questions. How can they quickly deploy the Symantec incorporates robust options for environment—hardware, hypervisors, and VMs—in managing both physical and virtual systems a standardized and automated way? How do they in a single tool, helping administrators know what they have and where it is? How do to seamlessly manage the layers of their they know what is broken or about to break? And virtualized environments and increase overall how can they fix what is broken? operational efficiency.

ndividual PCs do not use a lot of power—but for midsize organizations and large enterprises that have hundreds or thousands of end users, each I with his or her own computer, small reductions in power consumption can add up to annual savings that range from thousands to millions of dollars. In fact, the savings generated by reducing power consumption on individual PCs across a large enterprise can actually add up to more than the savings an organization can achieve by optimizing power management in its data center. Power management features are available on virtually every PC, but are not always set up and managed in Deploying automated a way that minimizes energy use without reducing user productivity. Dell KACE K1000 management plug-and-play appliances can help IT organizations efficiently manage power configurations to take advantage of power management these features, helping to reduce power consumption while maintaining productivity across the enterprise. for PC environments Small power management changes lead to big savings Automated PC power management helps Enhanced power management can accrue significant organizations make minor changes that can result savings and help IT staff efficiently control the in significant cost reduction by enabling them to standardize, implement, and enforce power usage ™ PC infrastructure. The Dell KACE K1000 policies throughout the PC infrastructure. These management appliance offers this capability measures can also help them meet significant energy-use reduction targets. in a flexible, easy-to-use package. In many locations, utilities rebates are available to organizations that implement, document, and enforce power management policies (see the “Antioch Unified School District flexes its power with Dell KACE appliances” sidebar in this article). This utility rebate alone can be a tremendous incentive for organizations to implement a power management strategy for their PC infrastructures. Power management techniques also help enterprises demonstrate environmental responsibility by reducing their carbon footprints and greenhouse gas emissions. In the public sector, many cities, counties, states, and federal government agencies are now actually required to implement energy-use reduction programs. And because heat can negatively affect PC reliability (which can consume help-desk and repair resources and keep employees from doing their jobs effectively), power management technologies can indirectly support high user productivity.

90 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. “We chose the Dell KACE [K1000 management] appliance to help us manage our PC power usage because it gave us the visibility and tools we needed to identify computers that were left on nights and weekends—wasting energy, costing us money, and taking resources out of our classrooms,” says Tim Forrester, executive director of operations for AUSD. “It also allows us to monitor computers remotely without disrupting our classrooms.” The AUSD IT team manages approximately 4,200 PCs scattered throughout the 26 schools in the district. “We encouraged teachers and students to turn off computers overnight and on weekends, but in the past, our staff had no way to know how many of these computers were actually left on,” says Bennie Driskell, supervisor of technology support for AUSD. Antioch Unified School “After we implemented the Dell KACE appliance, we discovered that about 800 computers were being left on overnight. Now District flexes its power that we’re using the auto-shutdown features of the appliance, with Dell KACE appliances we’re down to just 16 computers left on overnight.” In addition to the direct savings from reducing its power bills, Like many public school systems, Antioch Unified School District the district saves indirectly by reducing PC wear and tear and (AUSD) in California is always on the lookout for ways to reduce by cutting down on the heat generated by idling computers— its infrastructure maintenance costs and divert those funds to the thereby reducing the need for air-conditioning in its buildings classroom. And as a forward-thinking and environmentally aware during the hot summer months. What’s more, a US$61,500 organization, the district saw enhanced PC power management rebate from Pacific Gas and Electric virtually offset the cost as an opportunity to achieve two goals at once: save money and of the KACE management appliance—giving AUSD a nearly become more environmentally responsible. immediate return on its investment.

Dell KACE K1000 appliances streamline Microsoft® Windows®, Linux®, and Apple Mac OS X power management operating systems, enabling IT departments The Dell KACE K1000 management appliance helps to cost-effectively manage power usage organizations discover and inventory devices as well throughout their entire client infrastructure. For as create, deploy, and enforce power configurations. organizations running Windows XP, the appliances Reports on power usage throughout the PC provide comprehensive support for the power infrastructure helps administrators refine and optimize management settings in that OS. For organizations power management policies over time. running or planning a migration to the Microsoft Available either as hardware or as a VMware® Windows Vista® or Windows 7 operating systems, virtualization–ready virtual appliance, the Dell the appliances support the significantly expanded KACE K1000 appliance is designed to be simple power management features available in those to deploy—in fact, 68 percent of installations are operating systems—including enhanced granular completed in less than a week.1 Easy-to-use wizards power control and the ability to develop multiple allow administrators to control and enforce power profiles for different conditions. For example, settings remotely. an end user could use one power profile when Dell KACE K1000 appliances support displaying a presentation, and then use a different power management for computers running power profile when attending a meeting.

1 Based on KACE customer survey, November 2009.

Flexible policy management Power management policies must be specific to Power each type of end user. How quickly and how often management one user puts a system into a low-power state can best practices be very different from another. For example, some may use their PCs a few minutes each hour, and do Geared specifically for IT administrators seeking to not necessarily need instantaneous access to their implement a PC power systems. Power users, in contrast, may be on their management strategy PCs continuously, and need immediate access to across their organization, this Webcast explores key their systems at all times to remain productive. facets to consider before Dell KACE K1000 management appliances allow deployment as well as the tangible benefits such IT administrators to group users logically, through a strategy can provide. user-assigned labels (logical groupings), and then

kace.com/resources/ apply different power policies based on those labels. power-management This capability allows organizations to develop -webinar aggressive power policies that may, in a matter of Figure 1. Wizards help simplify managing power settings for a minutes, put a system into a sleep state for users who PC infrastructure use their systems only 10 minutes an hour or just a few times a day. At the same time, it enables far less Figure 1). With its wide range of systems management aggressive policies for users who are on and off their features and easy-to-use browser-based administration systems constantly. In this manner, IT administrators interfaces, the Dell KACE family of appliances gives IT can provide users with individualized service while still departments the ability to control and enforce power saving power enterprise-wide. management policies remotely.

Cost-effective deployment Efficient power management saves more Because Dell KACE management appliances are than utility costs designed to be both immediately productive as well as Multiplied across thousands of PCs, small reductions in trouble free for the long term, they can help deliver low power consumption can generate significant savings total cost of ownership. Plug-and-play deployment, one- in utility costs. But the benefits do not end there. The click software updates, and automated nightly backups automated power management capabilities available help to reduce installation and maintenance costs. in Dell KACE management appliances also help IT Also available as virtual appliances, Dell KACE organizations achieve other energy-use reduction goals management appliances can leverage an existing and demonstrate responsibility for the environment. virtualized server environment and thereby avoid the In addition, many power utility companies offer need for additional hardware in the data center. In rebates to organizations that implement, document, addition, organizations can reap potentially significant and enforce power management policies—which cost savings both directly (by putting PCs in a low- can further justify deploying Dell KACE management power state when they are not being used) and appliances for client computers. indirectly (by allowing the organization to save on heating and especially cooling). Learn more

Systems management cost calculator: Easy installation and utilization kace.com/cost-calculator Dell KACE management appliances are designed to

take the complexity out of PC power management. Dell KACE power management: A simple plug-and-play architecture helps virtually kace.com/power-management eliminate installation and maintenance, and the wizard through which IT administrators manage PCs Dell KACE green IT: provides a familiar, browser-style tabbed interface (see kace.com/green-it

Simplifying custom Microsoft Windows deployments

By Gong Wang and Wayne Liles

Dell™ Unified Server Configurator version 1.4 can help simplify and accelerate the deployment of customized Microsoft® Windows® operating systems on 11th-generation Dell PowerEdge™ servers.

he 11th-generation Dell PowerEdge Deploying Windows with standard media server family introduced the Lifecycle and a configuration set Controller embedded management In small organizations, Windows operating systems T solution, which is designed to eliminate are often deployed directly using standard Windows the media-based delivery of drivers, tools, and media, rather than from the network. To help ensure utilities necessary for systems management. One consistency across systems, administrators can of its key components, the Dell Unified Server customize the Windows installation by using the Configurator (USC) tool, can be used to configure Windows System Image Manager (Windows SIM) basic device settings, diagnose systems, and tool to create a configuration set. This configuration deploy selected operating systems by providing set comprises an answer file and a set of source the embedded device drivers for OS deployment. files, which can include out-of-box drivers, In addition to supporting OS deployment using applications, and packages such as security updates, standard OS media, USC 1.4 also supports hot fixes, and language packs. When using a customized media. configuration set, administrators can also automate The Microsoft Windows Server® 2008 and the OS deployment process in an unattended mode Windows Vista® operating systems, meanwhile, and avoid the installation prompts. introduced the Windows Automated Installation Figure 1 illustrates the high-level workflow when Kit (Windows AIK), a set of tools that assist with using Dell USC to install the OS with standard media the installation, customization, and deployment and a configuration set. Administrators use USC to of Windows operating systems. An updated and configure the RAID settings, create a disk volume, enhanced version of Windows AIK was later and, by selecting the desired OS, provide the latest released with the launch of Windows Server 2008 device drivers automatically at the beginning of R2 and Windows 7. the OS installation. This method helps greatly Administrators can take advantage of Dell USC simplify OS installation. to help simplify OS deployment, reduce deployment time, and eliminate potential problems caused Creating a configuration set by traditional media-based OS deployments. And To create a configuration set, administrators by using it in combination with key Windows AIK would first use the Windows SIM tool to create a tools, they can effectively and efficiently create catalog file—a binary file that lists the settings and and deploy customized Windows installations on packages within a Windows image. A catalog file 11th-generation PowerEdge servers. is automatically created when a Windows image

Create answer Copy answer file and file and configuration set to the configuration set root of a USB flash drive

Reboot system OS installation Boot system Create RAID Select OS Configure Yes with OS media and completes with Dell USC RAID? configuration to install USB flash drive unattended

Figure 1. Workflow for deploying Microsoft Windows using Dell USC with standard media and a configuration set

(.wim) file is opened in Windows SIM. Different versions installation by adding applications, packages, and of Windows SIM have different restrictions on creating out-of-box drivers to create the configuration set. catalog files for different Windows image architecture These files are hosted on the distribution share and types; the 32-bit version of Windows SIM is typically referenced in the answer file.2 recommended because it can create catalogs for all Finally, administrators should use Windows SIM supported Windows architecture types. to validate the settings in the answer file, save the After selecting a Windows image and creating answer file, and create the configuration set in a the catalog file, administrators can use Windows specified location. By default, the configuration set SIM to create a new answer file. An answer file is an includes an answer file named AutoUnattend.xml and XML-based file, typically named AutoUnattend.xml, a folder named AutoUnattend_Files, which includes that stores the custom settings used during Windows the distribution share content and its folder structure. Setup. To complete an unattended installation, the answer file must at a minimum specify the Deploying the OS Windows Setup language and location settings, Now that the configuration set is in place, administrators accept the software license terms, provide a product can store it on a USB flash drive and use that drive key, configure the disk partition, and provide the to carry out customized OS installations. As Figure 1 location of the Windows image file to install (either illustrates, USC can assist with the major steps of an by specifying the disk ID and partition ID or by installation on 11th-generation Dell PowerEdge servers; telling Windows Setup to install the OS on the first after administrators have set the RAID configuration available partition). Administrators can also specify (if necessary) and selected the OS, USC prepares the other settings to apply to the installation, such as drivers for the selected OS. When the system reboots user account names, display settings, and Microsoft from the Windows media, it automatically installs the Internet Explorer® Web browser settings.1 device drivers provided by USC, configures the Windows The next step is to create a distribution share— settings, and installs the applications, packages, and the folder that hosts the additional source files for a additional drivers provided by the configuration set. Windows deployment. This share can be set up on a Although using USC for OS deployment is local folder or on a shared network folder. The source not required, doing so helps ensure that the latest files can include third-party drivers, applications, and certified drivers are injected during the OS installation, Microsoft update packages such as security bulletins. and avoids the need to assemble the driver packages Administrators can create this folder using either and detail them in the configuration set. Before Windows Explorer or the Windows SIM tool. installation, administrators should be sure to verify After the distribution share and answer file are that the latest USC OS driver packages (available from in place, administrators would next customize the the Dell support Web site) are installed.

1 For more information on working with answer files, visit technet.microsoft.com/en-us/library/dd744318(ws.10).aspx. 2 For more information on adding applications and out-of-box drivers to configuration sets, visit technet.microsoft.com/en-us/library/dd799266(ws.10).aspx.

94 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Boot Conﬁgure Dell USC Reboot from Select Insert Perform system RAID or prepares inserted OS OS media validation Yes with Dell retain existing drivers for the media and begin to install media pass? USC conﬁguration selected OS OS installation

Continue No with inserted Yes media?

Figure 2. Workflow for deploying Microsoft Windows using Dell USC with customized media

Deploying Windows with customized media to continue after showing a confirmation prompt, Administrators can also merge a configuration set and still prepares the driver set for installation. with the standard Windows media by storing the configuration set directly on the media. To do so, Customizing a Windows image they would create a copy of the master image with offline servicing from the Windows DVD as well as a copy of the In addition to merging a configuration set with configuration set, use the Windows AIK oscdimg tool standard media, administrators can also use two to create a new ISO image containing these copies, other methods to customize the Windows installation burn this image to a DVD, and use Dell USC to deploy image with Windows AIK tools: offline servicing Windows from this customized media the same way and online servicing. Offline servicing enables the they would with standard media. Figure 2 illustrates injection of Microsoft update packages and hot fixes, the high-level workflow when using this method. updated drivers, setting changes, and language packs The advantage of this approach is that it does not into a Windows image before installation. Using this require a USB flash drive, avoiding a potential failure method helps administrators efficiently manage caused by inserting the drive into the non-primary set existing customized Windows images by avoiding the of USB ports for the system. However, one drawback need to deploy the image and then recapture it after is that it requires re-creating the customized media the changes are made. whenever the configuration set is updated to include Figure 3 shows the high-level workflow when newly released Microsoft security updates, change customizing a Windows image through offline settings, and so on. servicing. Typically, administrators would mount the Dell USC includes a verification mechanism to install.wim file, inject the required updates and make help ensure that the appropriate drivers are prepared the necessary modifications, and then unmount that for deployment. If the deployment uses nonstandard file. In Windows Server 2008 R2, Windows 7, and or customized media, USC 1.4 allows the installation the latest version of Windows AIK, the traditional

Add or remove drivers

Add or remove Mount Windows features Unmount Create customized install.wim image and ISO image Figure 3. Workflow for customizing a ﬁle using commit to using the Windows Microsoft Windows image with offline servicing DISM tool Add or remove changes AIK oscdimg tool language packs

Change Windows image to a higher edition

Build Windows Copy OS media Perform OS Preinstallation deployment with source ﬁles Environment to network share Dell USC (WinPE) media

Create answer Build reference Update online Prepare Use WinPE Replace install.wim file and installation image in image to capture image file with configuration set (OS customization) audit mode captured image

Create customized Figure 4. Workflow for customizing a Microsoft Windows image with online servicing ISO image using the Windows AIK oscdimg tool

image servicing tools (including pkgmgr, PEImg, and running the command sysprep /audit /reboot, IntlConfg) have been replaced by the Deployment or (for unattended installations) by setting the Image Servicing and Management (DISM) tool, Microsoft-Windows-Deployment\Reseal mode in the which administrators can use to mount and service answer file to audit. Windows images offline. The DISM tool does have some limitations, however: it can add only .cab Simplifying customized Windows files, .msu files, and .inf files offline. Adding system deployments component files such as .msi and .exe files, or When choosing a method for deploying customized adding service packs, requires using the online Windows operating systems, administrators should servicing method. first identify the requirements of their specific organization and make sure that the selected Customizing a Windows image method complies with the relevant license terms. with online servicing In addition, they should keep the Dell USC firmware Online servicing refers to the process of modifying up-to-date to help ensure that the latest certified Windows while the OS is running, and then capturing drivers are available. Following best practices the modified image for later deployment. This and using USC in conjunction with Windows AIK method allows administrators to install service packs, can enable administrators to perform consistent, update packages, system components, applications, repeatable OS deployments on 11th-generation drivers, and language packs on a running Windows Dell PowerEdge servers, helping minimize possible installation booted into audit mode. problems while simplifying and accelerating A key difference from offline servicing is that installations. applications can be installed and tested during this online process, which allows significantly increased Gong Wang is a software engineer in the Dell flexibility. The drawback, compared with offline Server Operating Systems Group. servicing, is that the image must be deployed, manipulated, and then recaptured. Both methods Wayne Liles is a systems management software strategist for the Dell Lifecycle Controller product. can be useful depending on the types of changes, updates, or additions that are required for a particular customization. Learn more Figure 4 shows the high-level workflow when

customizing a Windows image through online Dell embedded server management: servicing. Administrators can boot Windows into dell.com/embeddedservermanagement audit mode by pressing Ctrl+Shift+F3 from the Windows Welcome screen after the installation Dell PowerEdge servers: completes, by opening a command prompt and dell.com/poweredge

96 2010 Issue 03 | dell.com/powersolutions Reprinted from Dell Power Solutions, 2010 Issue 3. Copyright © 2010 Dell Inc. All rights reserved. Well-balanced for business The Dell™ PowerEdge™ R715 server: powered with up to 24 CPU cores and 256 GB of memory

With server technologies designed for energy efficiency without compromising performance, AMD and Dell help your business perform while meeting cost and power-usage goals. The PowerEdge R715 also includes power-management features such as programmable voltage regulators, power-regulating processors, and an interactive LCD screen for easy access to power-consumption information. Efficient, powerful, manageable—and a great choice for businesses of any size. Learn more at www.dell.com/poweredge/amd

© 2010 Advanced Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow logo, AMD Opteron, and combinations thereof are trademarks of Advanced Micro Devices, Inc. Other names are for informational purposes only and may be trademarks for their respective owners. Applications “Unlimited” Oracle – Siebel – PeopleSoft – JD Edwards

Continued New Releases

Customer Driven Product Roadmaps

Dedicated Development Teams

No Forced Migrations

Get better results with proven applications tailored for your industry, processes and geography.

oracle.com or call 1.800.ORACLE.1

PUB NOTE: Please use center marks to align page. CUSTOM Job No.: 001592 Live: 7.875” x 10.375” APPROVALS BY DATE PRODUCTION NOTES READER File Name: EBS_AppsUnlimited_1592_DPS Trim: 8.375” x 10.875” Traffic Product: Applications Bleed: 8.625” x 11.125” Production 01 8.375” x 10.875” Headline: Applications “Unlimited” Color: 4C LASER% Proofing Date: Fri, Jul. 23, 2010 1:45 PM Production: Graphic Mgr. Pub: Dell Power Solutions Dell Power Solutions Fonts: Traffic: HQ Adv. Mgr. Univers LT Std. Font Family Released Magazine Library Ref.: 000 Buddy Check Please examine these publication materials carefully. Any questions regarding the materials, please contact Darci Terlizzi (650) 506-9775