Cisco UCS
Download as PPTX, PDF7 likes11,391 views
The document outlines the Cisco Unified Computing System (UCS) and its various components, including rack and blade servers, management systems, and deployment applications. It emphasizes the benefits of UCS, such as improved space utilization, power efficiency, and reduced management points, while also addressing its weaknesses. The presentation provides insights into how UCS integrates networking and computing, optimizes virtualization, and enables dynamic resource provisioning.
Cisco UCS
- 1. Cisco Unified ComputingErik BussinkSenior Virtualization Specialist17/02/2011
- 2. AgendaCisco Unified Computing SystemHow we got here ?UCS IntroductionUCS ComponentsService ProfileUse CasesDemo
- 3. ServerServerServerServerServerServerServerServerServerServerServerServerServerServerServer Deployment: RackmountsFirst generationRack-optimized Top of Rack or End of Row switches CablesBenefitsSpace utilizationHighly flexibleWeaknessCablingServiceabilityPower efficiencyRackRack
- 6. Fixed backplane BenefitsSpace utilization
- 10. Large chassis needed to amortize switch/mgmt costsRackRack
- 11. ServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerServerManagementManagementManagementManagementServer Deployment ManagementChassis ManagementNew management layerBenefitsConsistency in chassis
- 12. Shared chassis infrastructure monitoring WeaknessAdditional mgmt overhead
- 14. Need chassis aggregation management
- 15. Artificial aggregation pointServerServerServerServerServerServerServerServerServerServerServerServerServerServerManagementManagementManagementManagementServer Deployment: ApplicationsApplication DeploymentSingle application per OSBenefitsIsolation
- 16. Flexibility
- 18. Power & Cooling
- 19. Server sprawlManagementManagementServer DeploymentSoftware SwitchSoftware SwitchSoftware Switch
- 20. Software SwitchSoftware SwitchSoftware SwitchManagementManagementManagementManagementServer Deployment Virtualization InfrastructureServers & Switches
- 21. Physical & VirtualChallengesMany points of management
- 23. Diagnostics
- 24. Training
- 25. SecurityMgmt ServerManagementManagementServer Deployment TodayOver the past 10 yearsAn evolution of size, not system
- 26. More servers & switches than ever
- 27. More switches per server
- 28. Management applied, not integratedResultMore points of management
- 29. More difficult to maintain policy coherence
- 30. More difficult to secure
- 31. More difficult to scaleMgmt ServerMgmt ServerMgmt ServerCisco Unified Computing SystemEmbed managementUnify fabricsOptimize virtualizationRemove unnecessary switches,adapters,management modulesLess than 1/3rd infrastructure
- 32. Mgmt ServerCisco Unified Computing SystemA single system that encompasses:Network: Unified fabric
- 33. Compute: Industry standard x86
- 35. Virtualization optimizedUnified management modelDynamic resource provisioningEfficient ScaleCisco network scale & services
- 36. Fewer servers with more memoryLower costFewer servers, switches, adapters, cables
- 38. Fewer points of managementBuilding Blocks
- 39. Overall System (Rear)Converged NetworkUplinksUplink: 8x 10Gb
- 41. Uplink: 2x 10Gb
- 43. Uplink: 4x 10Gb
- 44. Oversubscription 1:2MGMTSANSANLANSystem ComponentsFabric Interconnect(40 or 20 10GE ports) + (2 or 1 GEM slots)ChassisUpto 8 half width blades or 4 full width bladesFabric ExtenderHost to uplink traffic engineeringUp to 80Gb Flexible bandwidth allocationCompute BladeAdapterVirtualized adapter for single OS and hypervisor systemsFabricInterconnectFabricInterconnectSSGGGGAAGGGGGGCompute ChassisFabricExtenderFabricExtenderIICCRRx8x8x8x8Compute Blade(Half slot)Compute Blade(Full slot)AdapterAdapterAdapterPMPBBx86 Computerx86 ComputerXXXXXX
- 45. UCS Fabric Interconnect PortfolioUCS 6100 Family40-Port Fabric Interconnect40 fixed ports 10GE/FCoE, fixed
- 46. 2 Expansion Modules20-Port Fabric Interconnect20 fixed ports 10GE/FCoE, fixed
- 47. 1 Expansion ModuleExpansionModulesEthernet 6 Ports 10GE/FCoEFC + Ethernet 4 Ports 10GbE/FCoE
- 48. 4 Ports 1/2/4G FCFibre Channel 8 Ports 1/2/4G FCFibre Channel 6Ports 1/2/4/8GFCUCS 5108 Blade ChassisChassisUp to 8 half slot bladesUp to 4 full slot blades 4x power supplies, N+N grid redundant8x fans included2x UCS 2104 Fabric ExtenderAll items hot-pluggableUp to 40 chassis per UCS system
- 49. B-Series Family ComparisonUCS B230 M1 Compact -Performance Blade ServerUCS B200 M2 General PurposeBlade ServerUCS B440 M1 High-Performance Blade ServerUCS B250 M2Extended Memory Blade Server Compute & memory-intensive server for enterprise-critical workloadsMemory-intensive server for virtualized and large-data-set workloadsCompute & memory-intensive server for enterprise-critical workloadsHigh-density server with balanced compute performance and I/O flexibility
- 50. Three Pronged Adapter StrategyCostCompatibilityVirtualizationExisting Driver Stacks“Free” SAN Access for Any Ethernet Equipped HostVirtual Machine Aware: Virtualization and Consolidation “Palo”Converged network adapters (CNA)Ability to mix and match adapter types within a systemAutomatic discovery of component types
- 51. C-Series Product DetailsCisco’s Unified Computing System C-Series Rack-Mount servers now address the vast majority of data center compute requirements UCS C200 M2 High-Density Rack-Mount ServerUCS C210 M2General-Purpose Rack-Mount Server UCS C250 M2 Extended Memory Rack-Mount Server UCS C460 M1 High-Performance Rack-Mount Server High-density server with balanced compute performance and I/O flexibilityGeneral-purpose server for workloads requiring economical, high-capacity, internal storageHigh-performance, memory-intensive server for virtualized and large-data-set workloadsCompute & memory-intensive server for enterprise critical workloads
- 52. Unified Computing System ManagerEmbedded device manager for family of UCS componentsEnables stateless computing via Service ProfilesEfficient scale: Same effort for 1 to 320 bladesAPIs for integration with new and existing data center infrastructure
- 53. Service ProfilesA service profile is a logical representation of a server that specifies its:Identity—UUID, MAC, WWN, etc.
- 54. Configuration—Server Requirements, Boot Order, Firmware, etc.
- 55. Connectivity—VLAN, VSAN, QoS, etc.ServiceProfilePhysical ResourcesServiceProfileTemplatePoliciesLogical Resources
- 57. Server Mobility with Service Profiles Compute Node A is being upgraded.Service ProfileService Profile can be manually migrated to Compute Node B.
- 58. Logical Server MobilityTime AIdentityLAN/SAN ConfigurationService Profile: WebServerATime B
- 60. Server Upgrades:
- 61. Within a single Cisco UCS system
- 62. Across Cisco UCS systems
- 65. Disaster RecoveryServer AvailabilityToday’s deployment:Provisioned for peak capacity
- 66. Spare node per workloadWebVMwareOracle Blade Blade Blade Blade Blade Blade Blade Blade Blade Blade Blade BladeBurst capacity Blade Blade BladeHA spareWebVMwareOracleWith server profiles:Resources provisioned as needed
- 67. Same availability withfewer spares Blade Blade Blade Blade Blade Blade Blade Blade Blade Blade BladeVirtual Interface CardService ProfilesCisco UCS enablers:
- 68. Demo
- 69. Questions