Storage Solutions for Enterprise Consolidation with VMware
Peter C. Conway Vice President – Business & Product Operations Symmetrix Product Group EMC Agenda
• Overview • Business Continuity • Customer Case Study • Scalable Performance • Customer Case Study • Scaling VMware VDI
22 EMC’s Storage 2008 Portfolio CLARiiON CX4 UltraScale Series DMX-4 950 Symmetrix
CX4-960 NS80 NSX Flash drives Celerra CX4 CX4-240 CX4-480 -120 Flash NS40 drives Flash NS40G drives NS20 Fiber Channel and iSCSI NS80G DMX-4 NAS, FC, & iSCSI AX4 Virtual Provisioning EMC ControlCenter Consumer Storage SMC Navisphere Celerra Manager Connectrix SAN Connectivity Smarts PowerPath MPFSi RecoverPoint
LifeLine SRDF TimeFinder MirrorView SnapView
Mozy Remote Backup Replication Manager Replistor Xtender Family Powered by EMC Fortress RecoverPoint NetWorker Avamar CDP and CRR
DLm DL4106 4080 DL4206 Software-as- a-Service (SaaS) DL3D 3000 Invista 2.0 DL210 DL3D SAN Virtualization Centera 1500 EMC Centera Rainfinity Gen 4 LP Node DL4406 Rainfinity Global File Virtualization EMC Disk Library
33 Enterprise Customers Prefer Storage Area Networks*
SANs remain the preferred platform for virtualized environments by 2:1 margin and growing
• - 12 Months: 43% SAN Favorable vs. 29% for NAS • Today: 46% SAN Favorable vs. 25% for NAS • +12 Months: 59% SAN Favorable vs. 24% for NAS
“EMC remains well ahead of the pack as the most preferred storage vendor in virtualized environments, with IBM a distant second.”
*Goldman Sachs, IT Spending Survey , 08 September 2008- p. 11
44 Enterprise VMware Storage Requirements
• Virtualization drives physical consolidation • Server • Storage Tier 1 • Network Mission-critical applications • Consolidation requires 24 x 365
• Never lose data! Tier 2 Business-critical Business Continuity applications • Every application performs! Scalable Performance Tier 3 Business-support
VMware Deployments VMware applications
55 Symmetrix DMX-4 Tiered Storage Platforms The World’s Most Trusted Storage Platforms
Symmetrix DMX-4: World’s Largest High-End Storage Array
• 96 to 2,400 disks • 2 to 12 channel directors * • 2 to 8 4Gb disk directors * • Fibre Channel, Flash, iSCSI, Gigabit Ethernet, FICON • Up to 512 (256 usable) GB Global Memory
DMX-4 950: High-End Entry Point
• 32 to 360 disks Current • 2 or 4 front-end/back-end directors customers power • Fibre Channel, Flash, iSCSI, Gigabit Ethernet connectivity over 6000 VMs • Up to 128 (64 usable) GB Global Memory with a single Symmetrix
CAPACITY AND PERFORMANCE SCALABILITY *Combinations may be limited or restricted based on configuration
66 VMware + Symmetrix = High Availability VMware Cluster
APP Both #1 Node APP APP APP in each OS APP APP Fails OS respective OS OS market OS OS
VMware Symmetrix • Automatic restart of VM in case of • Non-disruptive upgrades and unplanned server failures operations • N-to-N failovers with VM's easily • Full component-level redundancy moved with hot-swappable replacements • Application management remains • Dual-ported disks and global-disk unchanged hot spares • VMFS shared storage allows hosts • Remote support and proactive call- to access and power-on VM's home capabilities
77 VMware and Multi-site Business Continuity Symmetrix Remote Data Facility (SRDF)
SRDF used in worlds largest mission critical environments Financial Industry, Telecom, Healthcare, Manufacturing, Retail, and Services SRDF with VMware Site Recovery Manager (SRM) Single button kicks off recovery sequence for protected VM's for fast restart SRDF integrated and validated with major applications supported with VMware Best practices with Oracle, SAP, and Microsoft
88 SRDF – Most Widely Deployed Disaster Restart Solution
APP APP APP APP OS OS OS OS Source Target SRDF links APP Target Source APP APP APP OS OS OS OS Source/Target Target/Source
Disk mirror in real time between Symmetrix systems Source to target architecture Supports bi-directional remote mirror operations Independent of host, OS, applications, and databases
99 VMware Site Recovery Manager (SRM) with SRDF
APP APP APP APP OS APP OS APP OS APP OS APP OS APP OS APP Production OS OS Disaster OS OS Data Center Recovery Site
SRDF Source Target
Disaster event VM’s restarted Site Recovery Manager initiates restart actions that includes having the SRDF Adapter Read/Write enable the SRDF Target
1010 Recovery Plan Testing with VMware Site Recovery Manager and TimeFinder
APP APP APP APP Disaster Production OS APP OS APP OS APP OS APP Recovery Site Data Center OS APP OS APP OS OS OS OS
SRDF Source Target
Activate/Split Time Finder
Testing of a recovery with zero downtime to the protected VM's in the protected site using TimeFinder in the recovery site
1111 VMware with SRDF Advanced 3-Site Solutions
Production Local/Regional Disaster Out-of-Region Disaster Data Center Recovery Site Recovery Site
APP APP APP APP APP APP OS APP OS APP OS APP OS APP OS APP OS APP OS APP OS APP OS APP OS OS OS OS OS OS
SRDF/A SRDF/S R1 R21 WAN R2
Site A Site B Site C Achievable no data loss solution with improved RTO
1212 King & Spalding Skip McDowell
Current Responsibilities: is the engineering lead for King & Spalding, responsible for server deployment and administration (virtual and physical) and overall data centers management. Past Experience: Prior to working at King & Spalding, Skip worked at an major pharmaceutical where he managed data center provisioning and deployment. Education: Skip has civil engineer degree from the University of North Carolina at Charlotte
1414 About Us
• King & Spalding is an international law firm founded in Atlanta, Georgia in 1885. • Currently represents more than 250 public companies including half of the Fortune 100. Clients include The Coca-Cola Company, Sprint Nextel, General Motors and Bank of America. • 800 lawyers (2000 employees) with offices in Abu Dhabi, Atlanta, Austin, Charlotte, Dubai, Frankfurt, Houston, London, New York, Riyadh, San Francisco, Silicon Valley and Washington D.C.
1515 The Environment 350 Virtual Machines 19 ESX Servers Both sides active Fail to alternate site
Brocade Brocade 48000 48000
SRDF/S Over IP
1616 Benefits
Data Center Consolidation • Decommissioned 100 physical production servers (265 more opportunities) • Replaced physical test lab consisting of 36 physical servers High Availability • Set up 2 geographically spanned clusters as part of Exchange 2007 migration • Fail -over in either direction • Exchange only today, • Looking at Site Recovery Manager for general case • Planning for possible three-site in the future Scalability • Use Dynamic Resource Scheduler to balance work across cluster • Distribute virtual machines across both data centers
1717 SCALABLE PERFORMANCE EMC Technology Leadership Analysts on: EMC’s Q1 2008 Symmetrix Flash Drive Announcement
“EMC is the first enter prise infrastructure pla yer to incorporate flash disk into their arrays, which should give them a huge performance advantage at the very sector of the market that always seems to need more and more. “If it creates as big a gap in real life transaction processing shops as it does on paper, this could ver y well be one of those killer advanta ges that only appear every 10 to 15 years.” -Steve Duplessie Senior Analyst, The Enterprise Strategy Group, Inc January 2008
1919 Ultra Performance via Enterprise-Class Flash Drives
First and Only High End Array to support Flash Drives Fully integrated with Symmetrix supporting software functionality Multi-year planning and development program Design specifications, design, integration, qualification
Technology optimized for enterprise environments Flash FC interface to front end RAM cache with many parallel read/write engines Drives Wear leveling algorithms and self-healing capabilities for reliability End -to -end ECC for data integrity Ideal for Low Latency, High Transaction Workloads Available for New and Existing Enables database acceleration Symmetrix DMX-4 Reduces batch processing windows Good for indices and frequently accessed tables
Flash Drives becoming mainstream
2020 Enterprise Flash Drives Shatter Performance Barriers
10x faster response time 100% Read Miss Workload 60
55
30x more IOPS per Drive 50
45 32% less power per GB 40 35 98% less power per I/O 30 25
20 15K FC Drives 15
~1 millisecond 10 Response Time Millisecond Millisecond Time Time Response Response Read-Miss 5 0 Response Time! 0 500 1000 1500 2000 2500 Workload - IOPs
2121 Developing Hotspots Hotspots Virtual Machines result from concentration APP APP APP APP APP of active data OS OS OS OS OS on a few disks
VMFS Normal Volume Active Data Inactive Data
2222 Virtual Provisioning
Virtual Machines Virtual Provisioning APP APP APP APP APP spreads active OS OS OS OS OS and inactive data evenly across all disks in pool VMFS Virtual Provisioned Volume Active Data Inactive Data
2323 Dynamic Cache Partitioning Test & Development Ordering System
APP APP APP APP APP APP APP APP APP APP OS OS OS OS OS OS OS OS OS OS
VMFS VMFS Internal App Cache CACHE Cache Data Order System Data
2424 FANNIE MAE Christopher Louis
Current Responsibilities: Christopher J. Lewis is Fannie Mae's vice president for Networks and End-User Systems, reporting to the senior vice president – Technology Infrastructure and Operations. Lewis leads the vision and strategy for the development and operations of Fannie Mae's server and storage infrastructure. He manages the staff in multiple data centers, including two in the Washington, DC metropolitan area, and one data center outside of this region.
Past Experience: Prior to joining Fannie Mae in November of 2005, Lewis was chief technology officer for Telstra Big Pond, Sydney, Australia, responsible for the oversight of all engineering functions at Big Pond, the consumer internet services unit of Telstra, Australia's national telecommunications service provider.
Previously, he was the chief technology officer with OzEmail Internet, an MCI Company, Sydney, Australia.
Education: Lewis has a master of business administration and bachelor of science in computer sciences from George Mason University.
2626 Current
Configuration
• 275 VMs • Mixed lifecycle environment • 47 ESX Hosts • 34TB SAN Storage Benefits • 10 Clusters • 2 Sites • Deployment times down • Decommissioned 200 Servers • Avg CPU up to 80% • Improved Availability
Learnings
• Storage Configuration • Optimize for swap • Redefined SLAs
2727 Future
Growth Virtual Machines • Reach 750 VMs by end of 2009 APP APP APP APP APP • Increase proportionately more Linux VMs • Virtualize desktops OS OS OSAPP OSAPP OSAPP APP APP OS OS OS OS OS
Growth
• Use SRDF+SRM to automate site fail-over • Decouple “swap” to improve over- subscription • Tiered storage solution
Highly efficient server operation with multi-site failover
2828 SCALING VMWARE VDI How VDI Impacts Infrastructure Requirements
Traditional Environment Consolidated Environment 1,000’s of Desk top systems distributed Thin Clients access data and through the enterprise applications over the network
Thin Clients
LAN/ WAN
Desktop VMs
Distributed data needs to be managed Centralized boot disk and user data Each user has there own “C” drive with boot disk and user data Allows IT to enforce corporate policies for what is on desk top IT supports/maintains remote systems Includes hardware compatibility and SW Simplified software management and updates hardware independence
3030 VMware on Symmetrix Scalability - VDI
VDI Test Lab • Based in Santa Clara • Symmetrix DMX-4 2500 • 12 Blade Frames
3131 Scalable VDI Test Lab Goal Validating VMware Virtual Desktop Infrastructure at scale
Key statistics 168 total servers in 12 chassis in 3 racks 1344 cores and 5.25TB of memory Symmetrix DMX with 224 Fibre Channel drives 95% Reduction in boot image size from TimeFinder Snap technology TimeFinder snapshots - 375-500 VM’s per base LUN
3232 10,000 user VDI Configuration
VDM - Manages connection from client to hosted desktop
3333 12 Virtual Infrastructure (VI) Clusters & 1 DMX-4 2500 = 10,277 Virtual Machines
3434 Enterprise Storage
• Virtualization is the infrastructure of IT’s future Fannie Mae King & Spalding • Enterprise infrastructure drives requirements Business Continuity Scalable Performance • EMC Symmetrix is “Enterprise Proven” Gold standard remote replication with SRDF Performance leadership including flash drives
3535 Symmetrix with VMware Solutions Guide Describes the details of integrating VMware infrastructure with Symmetrix storage
• Installation
• Configuration
• Use of Symmetrix Software with Look to see if ESX Server you’ve won a free copy http://www.EMC.com/collateral/h (pick up at VDI kiosk ardware/solution- in EMC Booth) overview/h2529-vmware-esx- svr-w-symmetrix-wp-ldv.pdf
3636 Shameless Partner Plug!
Keynote: Designing the Next-Generation Datacenter • Chad Sakac-EMC, Scott Davis-Vmware, Ed Bugnion-Cisco Sessions • Virtualization of Oracle Production Environments • VMware Lifecycle Manager – The Customer Experience • Site Recovery Manager In Concert with Storage Replication Optimizing the Network and Management Architecture for Virtual Environments • Remote Office Data Protection and De-Duplication for Vmware • VMware Management Essentials: Virtual Dependency Mapping, Automated Root Cause, and Compliance Analytics for VMware enabled Data Centers
3737 Q&A
Breakout Session # PO 3824 Pete Conway VP, Business Operations Symmetrix Product Group EMC September 16, 2008