Ethernet Fabrics and the Cloud: Avoid the Fog and Smog

Ethernet Fabrics and the Cloud: Avoid the Fog and Smog

Ethernet Fabrics and the Cloud: Avoid the Fog and Smog Dr. Steve Guendert Brocade Communications Dr. Casimer DeCusatis IBM Corporation February 7, 2013 Session 12735 Abstract This session will discuss Ethernet Fabrics: what they are, what their business and technical value is, and how to implement them as part of your cloud architecture including with System z. It will also dispel misconceptions to clear the smog and fog from the cloud. The focus will be on the Open Data Center Interoperable Network (ODIN) model. 2 Agenda-Overview • Introduction • A need for progress in data center network design • Data center network transformation • What is an Ethernet Fabric? • The Open Datacenter Interoperable Network (ODIN) • System z • Conclusion and questions. 3 The Need for Progress is Clear 30 percent 42 percent 85% Energy costs alone Worldwide, buildings In distributed computing represent about 30% of consume 42% of all 85% of computing an office building’s total electricity – up to 50% of capacity sits idle operating costs which is wasted 18% 20x 50%+ Anticipated annual Growth in density of More than half of our clients have increase in energy costs technology during this plans in place to build a new data decade. Energy costs center/network facilities as they higher than capital are out of power, cooling and/or outlay space Forecasted evolution of • Rapidly increasing demand for 10 Gbps Ethernet (IEEE, 2007) server connections • Transition to 10G is happening now and will be mainstream from 2012 • Broad deployment of 10GBaseT will simplify DC infrastructure • by easier server connectivity, while delivering bandwidth needed for heavy virtualization and IO intensive applications • Server Virtualization • To stop wastage of server CPU resources • Exploding East-West traffic volumes • to support multitier applications and high performance computing • Proliferation and mobility of Virtual Machines • to address fluctuating workload Forecasted evolution of • by on-demand starting, moving and (hopefully also) decommissioning VMs FibreChannel (Infornetics) • Increased complexity • Drives focus to maintaining the infrastructure • Rather than to adding business value by leveraging new infrastructure services Forecasted evolution of • Rapidly increasing demand for 10 Gbps Ethernet (IEEE, 2007) server connections • Transition to 10G is happening now and will be mainstream from 2012 • Broad deployment of 10GBaseT will simplify DC infrastructure • by easier server connectivity, while delivering bandwidth needed for heavy virtualization and IO intensive applications • Server Virtualization • To stop wastage of server CPU resources • Growing traction of LAN/SAN convergence • 70% of IT decision makers believe in it (Forrester, 2011) • Barriers are organizational rather than technological • Proliferation and mobility of Virtual Machines • to address fluctuating workload Forecasted evolution of • by on-demand starting, moving and (hopefully also) decommissioning VMs FibreChannel (Infornetics) • Increased complexity • Drives focus to maintaining the infrastructure • Rather than to adding business value by leveraging new infrastructure services How is the data center evolving ? Automated Integrated Platform Manager 9 & SDN stack Virtualized 6. Virtual Machine Storage Pool network state Virtualized Network Resources 7 automation Virtualized [Network Hypervisor] Compute Pool 7. Multi-tenant aware Virtual View Network Hypervisor Integrated Physical View 1 8.Self-contained 5 Optimized expandable Single, 3 infrastructure 6 Scalable 4 2 vSwitch vSwitch 1.Fabric managed 9.Platform Manager Fabric as a single switch & Software COMPUTE STORAGE COMPUTE STORAGE 2.Converged fabric Defined 8 Networking Stack Seamless 3.Scalable fabric Elasticity… 4.Flexible Bandwidth Rack 5.Optimized Traffic Data Center Network Transformation From networks to Ethernet fabrics • Timeframe: 1990s Business Agility • Focus: Improve connectivity, packet Cost Efficiency delivery • Historically 1 app:1 server; north-south traffic • Virtualization limited scalability • Traffic load strain • Increasing east-west traffic • STP: one path, narrow VM mobility • Complex, underutilized, rigid VIRTUALIZATION VIRTUALIZATION LAN SAN SERVICES ON DEMAND Flat 8 2/6/2013 Data Center Network Transformation From networks to Ethernet fabrics • Timeframe: 2000s Business Agility • Focus: Improve performance/app Cost Efficiency delivery • A more powerful, flatter network • Higher traffic, east-west, avoid congestion • Collapse layers to reduce complexity • High density, high bandwidth, wire speed VIRTUALIZATION VIRTUALIZATION Packet • Layer 2 challenges remain… Delivery LAN SAN LAN SAN VM VM Hierarchical 1990s SERVICES ON DEMAND Improve Connectivity Flat 9 Data Center Network Transformation From networks to Ethernet fabrics • Timeframe: 2010s Business Agility • Focus: Improve agility Cost Efficiency • Large, flat Layer 2, high speed, high availability • All paths active—no STP • Flexible topology • Ability to converge IP/storage • Wide, intelligent Virtual Machine (VM) mobility • Manage as a single entity Application • Virtualize for the cloud Delivery VIRTUALIZATION VIRTUALIZATION Packet LAN Delivery SAN Private Cloud VM VM Flat LAN LAN SAN VM SAN VM VM VM Hierarchical SERVICES ON DEMAND 1990s 2000s Improve Improve Connectivity Performance Ethernet Fabric 10 Data Center Network Transformation From networks to Ethernet fabrics • Timeframe: 2015+ Business Agility • Focus: Improve the user experience Cost Efficiency • Leverage resources across data centers Service • More flexibility to scale Delivery • Relocate applications for greater efficiency • Layer 2 over distance, seamless mobility, rapid access • Building on expertise to extend the LAN Application VM SAN private cloud VM Delivery VM VIRTUALIZATION VIRTUALIZATION VM Packet Delivery LAN Fabrics SAN Extended VM Private Cloud LAN VM VM SAN VM VM Private VM LAN Flat Cloud VM SAN VM Data Center 2 LAN VM SAN Hierarchical VM SERVICES ON DEMAND VM VM 1990s 2000s 2010s Improve Improve Improve Agility Connectivity Performance Data Center 1 Fabrics 11 Data Center Network Transformation From networks to Ethernet fabrics Business Agility Cost Efficiency • Timeframe: 2015+ Orchestration • Focus: Orchestration Participation LAN VM SAN VM • Leverage service provider VM VM VM resources VM Service • Meet spikes/seasonal demand cost Data Center 2 Delivery LAN effectively VM SAN VM VM • Accelerate application deployment VM Fabrics • Resiliency in the event of a site outage Data Center 1 VIRTUALIZATION VIRTUALIZATION Application • Standards-based, open support, Delivery LAN VM SAN integrated management VM VM VM Extended Packet Private Delivery Cloud LAN SAN Hybrid Fabrics VM Cloud LAN VM VM SAN VM VM LAN Private VM SAN VM Flat Cloud VM Data Center 2 LAN VM VM SERVICES ON DEMAND SAN VM Hierarchical VM VM Public VM VM VM Cloud 1990s 2000s 2010s 2015+ Improve Improve Improve Agility Improve Connectivity Performance User Experience Data Center 1 Fabrics 12 Data Center Network Transformation Business Agility From networks to Ethernet fabrics Cost Efficiency Orchestration Participation LAN V M SAN V M V M V M VM VM LAN Data Center 2 Service VM SAN VM Delivery VM LAN V VM SAN V M M VM V V Public M M V V Cloud VM M M V V M M Fabrics Data Center 2 VIRTUALIZATION VIRTUALIZATION Application Packet Delivery LAN Data Center 1 VM SAN Hybrid Delivery VM Fabrics VM Cloud LAN VM VM SAN VM VM VM Data Center 1 LAN Extended SAN Private VM Fabrics Cloud LAN VM SAN Private Flat Cloud Hierarchical SERVICES ON DEMAND 2015+ 1990s 2000s 2010s Improve User Experience Improve Connectivity Improve Performance Improve Agility 13 What Are the Effects of This Transformation? Applications will be disaggregated Application Application Component Component Database Firewall Application Database DISTRIBUTED “By Next2014,-generation 80 percent data of networking centers will traffic need to will bechange between in an servers.” unprecedented fashion. –Gartner 2/6/2013 ETHERNET FABRICS Foundation for the Cloud User Benefits Cloud Business Benefits Quicker response to: Increased: • Needs Shared pool of • Business agility • Fiscal responsibility • Requests resources that can be dynamically allocated to users • Concerns Server Virtualization Pools of Compute and Storage Resources Dedicated to Applications Ethernet Fabrics A Network That Dynamically Meets the Needs of Applications Effortless Connectivity Better Service Delivery . Resilient . Flexible topology WHY . Scalable/elastic ETHERNET . Flat architecture FABRICS? Network Automation Simpler Service Orchestration Future-Proof Data . Logical chassis Center Networks . Automatic VM alignment . Seamless convergence of storage, voice, and video Enables organizations to: • Leverage IT as an asset • Reduce operational expenditures THE BUSINESS for BENEFITS OF data centers • Install a data center infrastructure ETHERNET that is transparent to applications and users because FABRICS it “just works” and is automated, flexible, and dynamic STANDARDS, TERMS, AND TECHNOLOGIES TRILL, SPB, Flat Networks, and Convergence Ethernet Fabrics 101 Vernacular Useful terms and definitions • TRILL (Transparent Interconnect of Lots of Links) and SPB (Shortest Path Bridging)—Standards that provide multi-path, multi-hop capabilities in Ethernet fabrics • Convergence—The ability of a single network infrastructure to support the needs of multiple technologies • Fabric-based infrastructure versus storage fabric versus Ethernet fabric: • Fabric-based infrastructure—A Gartner term that refers

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    46 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us