Principled Technologies Complexity and Cost Comparison Cisco UCS vs. IBM Flex System
COMPLEXITY AND COST COMPARISON: CISCO UCS VS. IBM FLEX SYSTEMMAY 2013A PRINCIPLED TECHNOLOGIES TEST REPORTCommissioned by Cisco Systems, Inc.Not all IT architectures are created equal. Whether you are updating yourexisting infrastructure or building from the ground up, choosing a solution that easesdeployment and streamlines management while keeping costs down is a wise choice.Cisco Unified Computing System (UCS®) and IBM Flex System both offer tools tostreamline deployment and management of your IT infrastructure, but which does moreto help you save in deployment, management, and hardware costs? We evaluated thetechnical features that the Cisco UCS and IBM Flex System architectures offer, andfound that the Cisco UCS Unified Fabric architecture and the Cisco UCS Manager helpalleviate deployment and management burdens with less hardware and withoutadditional software licensing. Additionally, the advantages of the Cisco UCS UnifiedFabric are not limited to the chassis but extend to rack servers and even virtualmachines. We also compared the costs of purchasing these solutions in different sizeddeployments, and found that the Cisco UCS solution could potentially reduce yourcapital expenditure by as much as 30.7 percent.
A Principled Technologies test report 2Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemSIMPLIFIED ARCHITECTUREThe Cisco UCS solution provides all management and configuration services atthe centrally located Fabric Interconnects, so you can manage large-scale deploymentsfrom a single location. This method lets you consolidate hardware and streamlinemanagement. The IBM Flex System solution uses a distributed management model withchassis-level control. This method adds complexity to the hardware configuration, whichcan increase management needs.As Figures 1 and 2 show, the converged network with Cisco UCS architecture isconsiderably less complex than the distributed network with the IBM Flex System.Figure 1: Enterprise view ofthe Cisco UCS managementsolution.Cisco UCS solution networkingFigure 2: Enterprise view ofthe IBM Flex Systemmanagement solution.(Note: Second FSM node isgreyed out due to lack ofsupport in the current releaseof Flex System Manager.Support is planned for a futurerelease.)IBM Flex System solution networking
A Principled Technologies test report 3Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemCisco UCS architectureThe Cisco UCS model features a converged fabric where all systemsmanagement and configuration originates from a pair of redundant Layer 2 networkswitches called Fabric Interconnects (FI). The Fabric Interconnect is a convergence anddistribution point that combines traffic from SAN, management, and data networks intoa single converged network that connects directly to the managed compute nodes.You can manage your entire Cisco UCS infrastructure from anywhere with theCisco UCS Manager, which is a highly available, single location for systems management.As your network scales out and you require multiple Layer 2 Fabric Interconnects, CiscoUCS Central consolidates all management systems together into one dashboard. A pairof Fabric Extenders (FEX) connects each Cisco blade chassis to the Fabric Interconnects.The FEX is not a switch; it simply aggregates FCoE connectivity without requiring anyuser configuration. With UCS Manager, you can apply all firmware versions and updatesto existing servers, new nodes, and other hardware with the click of a mouse.IBM Flex System architectureIn contrast, IBM Flex System architecture performs datacenter networking andmanagement at the blade chassis level. This requires separate connections for LAN,SAN, and management networks to each blade chassis. With IBM Flex System, there isno central management location for updating and configuring the compute nodes inyour datacenter. This means that every set of four blade chassis requires a separate FlexSystem Manager™ (FSM) appliance that you must log directly into to perform systemupdates. IBM currently provides no HA capabilities for the FSM, but has discussed aredundant option in documentation that would require a second standby “redundantpassive” FSM appliance.1Support for a second FSM appliance is planned for a futurerelease of Flex System Manager.Each IBM blade chassis also has its own set of up to four Layer 2 switches andtwo Chassis Management Modules (CMMs) that you must configure and update as well.Due to the IBM solution’s lacking a true converged network, half of the switch portsmust be dedicated Fibre Channel switches and the remaining ports must be Ethernetcapable. The increased number of cables, ports, IP addresses, and appliances to managecan add a significant amount of administration time for your IT staff.2Theseinterdependencies also increase risk of error when you make changes to yourinfrastructure. In our IBM Flex System configuration, we chose two base CN4093switches with the Switch Upgrade 2 license package to match the eight convergedEthernet and Fibre Channel connections on the Cisco UCS blade chassis.1http://www.redbooks.ibm.com/abstracts/tips0863.html#fsm2http://publib.boulder.ibm.com/infocenter/flexsys/information/index.jsp?topic=%2Fcom.ibm.acc.8731.doc%2Fwhats_new_120.html
A Principled Technologies test report 4Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemManaging your entire infrastructure with one solutionServer workloads can vary based on their requirements. For example, some arecompute intensive while others require large storage capacity and I/O. Because of thesedifferences, most server deployments contain a mix of both blade and rack mountservers. Ideally, the management solution you choose lets you support and manageboth types of hardware from a single interface. Solutions without this support createextra work for IT staff, who must perform every maintenance task (from firmwareupdates to server status monitoring) twice, using two separate tools. Maintaining twoseparate management software products also increases licensing costs.Cisco UCS Manager offers support for all current and previous generation UCShardware models, including both blade and rack servers. Regardless of the hardware inyour UCS deployment, you can manage it the same way from one unified interface. Thislets you optimize your infrastructure to match your workloads without sacrificing themanagement capabilities of UCS Manager or adding additional management products.This can make your infrastructure easier to manage and can reduce licensing costs.The current release of IBM Flex System Manager, version 1.2, supports only thecurrent generation of Flex System blade servers.3It does not support any of the BladeCenter series or rack-based servers. This lack of support can create additional overheadand the need for additional IT staff time when deploying and maintain the IBM FlexSystem solution.STREAMLINED DEPLOYMENTAs Figure 3 shows, updating and configuring a new, fully populated blade chassisin the datacenter is less complex using Cisco UCS architecture. This includes every stepfrom plugging in the chassis to when the servers are ready to begin productive work.Deploying the Cisco UCS Deploying the IBM Flex System1. Connect the power and network cables to the bladechassis.2. The hardware is auto-discovered by UCS Manager.Using one global Service Profile you can apply allfirmware updates and configure server BIOS andnetwork settings.3. Apply FEX module firmware update through UCSManager.1. Connect the power and network cables to the bladechassis.2. Connect to the CMM Web interface to create useraccounts and configure IP addresses for CMMs.3. Power on Flex System Manager node and connect toWeb interface.4. Follow FSM initial setup wizard and update managementsoftware and firmware. This could take up to two hours.45. If there are more than four blade chassis being deployed,3http://www.redbooks.ibm.com/abstracts/tips0863.html4http://publib.boulder.ibm.com/infocenter/flexsys/information/index.jsp?topic=%2Fcom.ibm.acc.8731.doc%2Fgetting_started.html
A Principled Technologies test report 5Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemDeploying the Cisco UCS Deploying the IBM Flex System4. Deploy OS to compute nodes using PXE boot andeither vSphere Auto Deploy or Orchestrator, MicrosoftDeployment services, or Red Hat Satellite services.you need to repeat steps 2 through 4 for each set of fourchassis.6. Using the FSM Web interface, you must “discover” andthen “inventory” every compute node in your deploymentenvironment before applying updates.7. Use FSM Configuration Patterns to configure computenodes from Flex System Manager interface. ConfigurationPatterns can also set up blade network switches ifadditional IBM Fabric Manager software is purchased.8. Deploy OS to compute nodes through the followingmethods:a. ESXi, KVM, and RHEL install via FSM image deployb. Windows Server by PXE Microsoft DeploymentservicesFigure 3: Deployment process comparison for the two solutions.The additional hardware of an IBM Flex System distributed networkmanagement solution demands not only additional cost, but also extra time for setupand configuration of each module, which grows with deployment size.Reducing network complexityIP addresses and switch ports increase administrator work and add extra cost innetwork cables and switches. The Cisco UCS solution reduces network complexity andcost by requiring only one converged network, while the IBM solution requires thatthree separate managed networks connect to each blade chassis (see Figure 4). The IBMmanagement network alone requires two extra cables and IP addresses in every chassisfor the CMM modules (still required even with FSM) along with cabling and IP for theFSM modules required for multi-chassis control. The Cisco UCS converged networkmakes it possible to send management traffic to the blades without burdening systemadministrators with any additional network management workload or cost overhead.
A Principled Technologies test report 6Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemFigure 4: Rear view of theCisco UCS and IBM Flex Systemblade chassis showing therequired network connections.(Note: Second FSM node isgreyed out due to lack ofsupport in the current releaseof Flex System Manager.Support is planned for a futurerelease.)HIGHLY AVAILABLE AND SCALABLE MANAGEMENTDowntime due to hardware outage or failure can be costly, so it is importantthat management solutions provide redundancy and make it easy for administrators totransition workloads to keep business moving. Again, Cisco UCS Manager provides aredundant solution more efficiently than IBM Flex System. One pair of Cisco FabricInterconnects automatically provides fault tolerant management for up to 160 servers,while similar hardware with IBM Flex System requires a labor-intensive failover andsetup process for its active/standby FSM nodes, is still awaiting official support, and canmanage only 54 servers. Further, Cisco UCS Central software provides centralizedmanagement at no charge for up to five domains (800 servers), which can helpadministrators get workloads moved and running should a failure occur. Figure 5compares how the two solutions deliver high availability.Cisco UCS Manager IBM Flex System ManagerUCS Manager is a centralized and model-based XML APIthat is both comprehensive and adaptive to changes in theenvironmentFlex System Manager is pre-packaged softwarerunning on a locked appliance blade pushing downscripted commands to HWRedundant Fabric Interconnects provide management to160 servers (blade or rack) per domainFlex System Manager nodes can manage only fourchassis or up to 55 blades, and there is currently noredundant FSM option availableAccess from a single-cluster IP address; Out of the boxfault tolerant with Active-Standby Unified Managementand Active-Active Unified Fabric for dataAccess each node by individual IP address; if a nodefails, log into secondary device IP addressManage multiple domains from a single interface with UCSCentral software; UCS Dashboard is free open-sourcemonitor of managersThere is no multi-FSM aggregation tool availablefrom IBM. Each FSM acts as an isolated node, andeach FSM maintains a separate software image toback up and archiveFigure 5: High availability comparison for the two solutions.
A Principled Technologies test report 7Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemAutomatic network failoverCareful planning and management is required to maintain a fully redundantnetwork. Due to its unified fabric model, Cisco UCS is the only solution that deliversautomatic network failover right out of the box – you do not need to configure anyhardware or software or make any additional purchases. A Cisco UCS network adapter isconnected to both fabrics (EtherChannel groups A and B), so even if one network pathfails, the host continues without interruption. Further, if an entire Fabric fails, allnetwork traffic routes to the second fabric automatically, without loss of connection.The IBM Flex System has no foundational architectural advantages like the CiscoUCS fabric architecture and requires extra setup and configuration to achieve equivalentfailover functionality. IBM Flex System requires you set up and maintain traditionalnetwork switches for each chassis. For every new chassis, you are required to configurethings such as load balancing, link aggregation, and VLANs, and carefully ensure thatchanges to an environment do not cause unintended consequences such as spanningtree loops and misconfigured ports. This adds another layer of management complexityand introduces risk, making management and change control more difficult.Centralized redundant management interface for large enterprisesBecause IBM Flex System Manager nodes do not failover automatically like theCisco UCS solution, administrators must manually connect to a backup node and bring itonline. Each target system has an OS agent that remains registered to the original FSMnode and does not recognize the new FSM. Admins must manually unregister each ofthese agents from the failed node and then register them to the new FSM node. If youdo not follow this procedure, you may get Request Access failures when you attempt tomanage these resources.5Similar to initial setup, each compute node needs to gothrough the discovery and inventory process again before the new FSM can managethem. Reconfiguring is an unnecessary hassle, and it can be time consuming to switchover to the backup management node in the IBM Flex System solution when comparedto the Cisco UCS fault-tolerant cluster, which does this automatically in a matter ofseconds.EXTENDS MANAGEMENT AND FABRIC TO RACK SERVERS AND VMSConsider the features available to you when you receive your hardware andwhat features you need to pay extra for. Hidden license fees and other support costscan reduce your overall return on investment and increase the time it takes to realizeactual gains from your purchase. There are many advanced features available for free5http://publib.boulder.ibm.com/infocenter/flexsys/information/index.jsp?topic=%2Fcom.ibm.acc.8731.doc%2Ftroubleshooting_the_software.htmlDue to its unified fabricmodel, Cisco UCS is theonly solution thatdelivers automaticnetwork failover rightout of the box.
A Principled Technologies test report 8Complexity and cost comparison: Cisco UCS vs. IBM Flex Systemwith the Cisco UCS solution that are not available or require additional licensing withIBM Flex System.Scale up without added management costsCisco UCS facilitates large-scale deployment and management with noadditional hardware or software requirements. Cisco UCS Manager and UCS Central areincluded at no additional cost with the purchase of a pair of Cisco Fabric Interconnects.Because Cisco Fabric Interconnects can replace top-of-rack (TOR) switches, theadvanced management features of UCS Manager are essentially free when compared toa traditional blade deployment. In addition, UCS Manager manages up to 20 chassis, andwith UCS Central scales up to 100 chassis (800 servers), with no additional cost, and10,000+ with nominal licensing.IBM Flex System Manager requires individual software licenses and supportcontracts for each Flex System Manager node, as well as additional license fees for eachblade chassis that the FSM nodes manage. Because a non-redundant FSM nodemanages only four chassis (54 servers), these costs can quickly add up. For example, 10IBM Flex System blade chassis would require customers to purchase 6 FSM appliancesand would require 10 Flex System Manager software licenses, just to manage thehardware with an active/standby management solution. Cisco UCS Manager providesthis functionality with true high availability at no charge. In addition, because the CiscoUCS solution scales up much more easily than the IBM Flex System, you are not requiredto overprovision by purchasing as much hardware up front. You can add compute nodesand networking as you need them.Advanced network policies and adapter settingsTighter integration of physical networks into the virtual world and features suchas Quality of Service (QoS) provide real value to datacenters. QoS lets networkadministrators guarantee a certain level of network bandwidth to VMs and physicalmachines. Some events, such as failovers, can use all of the available bandwidth in achassis. With QoS, administrators set a desired minimum bandwidth and the systemautomatically limits the other links to maintain server performance. Cisco provides QoSat no charge as part of its network offering. IBM requires you purchase a Flex SystemManager Service Fabric Provisioning software and support upgrade to unlock QoSfeatures from its switches.Cisco Unified Fabric and UCS Manager, extended beyond the blade chassisIn addition to the chassis, UCS offers the benefits of the unified fabric to rackservers as well. Cisco offers a complete line of Intel processor-based servers designed toCisco provides QoS atno charge as part of itsnetwork offering. IBMrequires you purchasean FSM Service FabricProvisioning upgrade tounlock QoS featuresfrom its switches.Cisco UCS Central canexist on a virtualmachine (VM) in ahighly available clusterto provide maximumuptime and resiliency ofits network IP address.
A Principled Technologies test report 9Complexity and cost comparison: Cisco UCS vs. IBM Flex Systemmeet a variety of business needs. All Cisco UCS C series servers integrate directly intothe Cisco Unified Computing Fabric. You can manage rack-mounted servers within thesame UCS Manager as your blade servers.Cisco UCS offers the VM-FEX feature to provide additional performance andsecurity to virtual environments. Similar to the Fabric Extenders (FEX), the VM-FEXfeature extends the management capabilities of the Fabric Interconnect switches intothe vSwitch and vNIC connections inside VMware® vSphere®.6The advanced securityand performance features Fabric Interconnects provide, such as QoS and EtherChannelredundancy, and VMware Direct Path I/O are available inside the hypervisor to provideVMs a direct link to the Fabric Interconnect. Cisco offers this feature at no charge withits UCS products; the IBM Flex System solution offers no similar feature.Cisco Service Profiles and IBM Configuration PatternsCisco UCS and IBM Flex System architectures both offer automation support tostreamline many common server setup tasks and keep them running smoothly. Bothsolutions provide an automation method for applying settings to network-connectedhardware and updating firmware: Cisco UCS uses Service Profiles, and IBM Flex Systemuses Configuration Patterns. While both solutions provide similar functionality, themethods to achieve them, the features they provide, and the associated costs are verydifferent.Firmware updates and management for individual nodesCisco UCS Service Profiles provide a one-stop shop for all of your organization’shardware setup and maintenance needs. In one Service Profile, you can set all of theBIOS, device, and firmware configuration settings for a compute node as well as updatethe firmware version. Service Profiles let you archive a backup firmware version for eachdevice to roll back to in case of error, with the click of a button. When Cisco UCSManager automatically discovers a new chassis, you can immediately run the ServiceProfile to configure the server and update its firmware. This is the advantage of the UCSdesign as a model-based unified management engine.When you connect a new blade chassis with IBM Flex System ConfigurationPatterns, you must first log into the CMM and configure it, and then log into the FSMand configure that as well before working with the blades. Flex System Manager mustdiscover and inventory each blade individually, which can be a time consuming process,before applying Configuration Patterns.Once an IBM server registers inside Flex System Manager, the ConfigurationPatterns apply to the compute nodes. The BIOS and device configuration settings apply6http://www.youtube.com/watch?v=8uCU9ghxJKgThe global ServerProfile approach ofCisco UCS Managerapplies firmwareupdates at once, to allsystems; the IBM FlexSystem solutionrequires individualupdates.
A Principled Technologies test report 10Complexity and cost comparison: Cisco UCS vs. IBM Flex Systemto compute nodes within the control of the FSM, which is limited to four blade chassisor 55 servers (54+FSM) in each domain. To apply Configuration Patterns to additionalservers, you must start over by creating entirely new Configuration Patterns that aredeployed from a separate FSM domain entirely. This process is repeated as you deployeach rack as a stand-alone domain.Unlike UCS, Flex System Manager updates firmware from a separate area, notfrom Configuration Patterns for your blades. With IBM Flex System, you must applyfirmware updates to the target systems using separate tools, four chassis at a time,which can take considerable staff time. The global Service Profile approach of Cisco UCSManager applies these updates at once, to all systems.Support for multiple server models in one profileMost infrastructures use multiple server models. The Cisco UCS solution cancombine different server models (both blade and rack) and configurations into oneglobal Service Profile to apply to an entire infrastructure in one update. This is a result ofUCS Manager being adaptive and model-based. Alternatively, the IBM Flex SystemManager is top-down software and does not support multiple models per ConfigurationPattern. This requires that you create and run a separate Configuration Pattern for eachserver model you deploy (i.e., x240 pattern only can apply to x240). Archiving,managing, and individually running different Configuration Patterns with the IBM FlexSystem solution increases the time and effort for IT administrators to keep the networkrunning. This leaves less time available for strategic projects that can provide additionalvalue to your organization.COST SAVINGS THAT SCALEIn this section, we compare the actual cost of hardware, software, andwarranties for the Cisco and IBM solutions. We picked configurations that providemaximum bandwidth and the high availability features we discussed in this paper. Wecompared the cost of three typical deployment sizes that a business might choose: 12blades, 40 blades, and 80 blades. We found that the Cisco UCS solution required lesscapital expenditure than the IBM Flex System solution in every deployment size weanalyzed. Savings from management and deployment efficiencies increase thisadvantage as the solution scales. For detailed costs, see Appendix A.Twelve-blade deployment (small size)First, we compare a small deployment of 12 blades and the required hardwarefor a fully functional solution. The Cisco UCS solution uses two blade chassis to equal thesame number of blades in one IBM chassis. (The second blade chassis is not fullypopulated, but we chose a blade number that would give the IBM solution the bestThe Cisco UCS solutioncan combine differentserver models andconfigurations into oneglobal Service Profile toapply to an entireinfrastructure in oneupdate.We found that the CiscoUCS solution requiredless capital expenditurethan the IBM FlexSystem solution inevery deployment sizewe analyzed. Savingsfrom management anddeployment efficiencieswould increase thisadvantage.
A Principled Technologies test report 11Complexity and cost comparison: Cisco UCS vs. IBM Flex Systempossible pricing.) Each UCS blade chassis has two Fabric Extender modules that connectto a pair of Fabric Interconnect switches.The IBM solution uses one fully populated blade chassis with 12 blade servers,two Flex System Manager nodes, and two CMMs. Although there is no automaticfailover supported currently with this configuration, it is planned for a future release, sowe chose to include two FSM modules in active/standby to emulate the redundancy ofthe Cisco UCS management solution. When IBM launches FSM failover support, theremay be additional licensing fees required, but none have been disclosed at this time.Because the IBM blade chassis contains two Layer 2 switches, we do not include a top-of-rack switch for the small deployment comparison. Figure 6 depicts 12-bladeconfigurations for both solutions.Figure 6: Side-by-sidecomparison of a typical small-sized deployment (12 blades).As Figure 7 shows, we found that the Cisco UCS solution cost 20.6 percent lessthan the IBM Flex System solution for the 12-blade deployment we analyzed.7This is allwhile delivering added functionality to ease deployment and infrastructuremanagement.7All costs in this study are from the PC Connection and IBM Web sites on 05/23/2013. We do not include discounts, tax, or shipping.
A Principled Technologies test report 12Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemFigure 7: A 12-bladedeployment using the Cisco UCSsolution can cost up to 20.6percent less than using IBMFlex System.Forty-blade deployment (medium size)In the 40-blade deployment, each solution can provide a set of fully populatedblade chassis. The Cisco UCS solution retains the pair of Fabric Interconnects and usesfive blade chassis with two Fabric Extenders in each.The 40-blade IBM solution contains three blade chassis, each with two CMMsand two blade switches. The pair of FSM nodes can still manage this deployment. Weinclude a pair of IBM RackSwitch G8264R switches to support the increased number ofnetwork connections required by the IBM solution. Figure 8 depicts 40-bladeconfigurations for both solutions.Figure 8: Side-by-sidecomparison of a typical mid-sized deployment (40 blades).$250,792.85$315,711.00$0$50,000$100,000$150,000$200,000$250,000$300,000$350,000Cisco UCS IBM Flex SystemCostinUSdollars12-blade deployment costs
A Principled Technologies test report 13Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemWe found that a 40-blade deployment of Cisco UCS infrastructure could cost upto 30.5 percent less than an equivalent IBM Flex System solution (see Figure 9).Figure 9: A 40-bladedeployment using the Cisco UCSsolution can cost up to 30.5percent less than using IBMFlex System.Eighty-blade deployment (large size)The large deployment replaces Cisco 6248UP Fabric Interconnect switches withCisco 6296UP Fabric Interconnect switches to support the additional connections thatthe Cisco blade chassis requires. The IBM Flex System solution requires two extra FSMnodes and licensing because it crosses the four-chassis threshold that one pair cansupport. Figure 10 depicts 80-blade configurations for both solutions.Figure 10: Side-by-sidecomparison of a typical large-sized deployment (80 blades).$708,935.99$1,020,209.00$0$200,000$400,000$600,000$800,000$1,000,000$1,200,000Cisco UCS IBM Flex SystemCostinUSdollars40-blade deployment costs
A Principled Technologies test report 14Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemIn a large deployment of 80 blade servers, the Cisco UCS solution can cost up to30.7 percent less to implement than its IBM Flex System equivalent (see Figure 11).Figure 11: An 80-bladedeployment using the CiscoUCS solution can cost up to30.7 percent less than usingIBM Flex System.IN CONCLUSIONMoving to a managed solution streamlines server deployment and reducesmaintenance time. Your infrastructure should be highly available, easy to use, scalable,and cost-effective to implement. Cisco UCS Manager provides a streamlined method forautomating hardware setup and firmware updates in one highly available solution tokeep management costs down. Due to its converged network model, Cisco UCSManager provides all of this functionality in a cost-effective package—saving up to 30.7percent—with no hidden fees or additional licensing costs. In contrast, IBM Flex Systemprovides fewer vital features out-of-box, increases network and managementcomplexity, requires additional hardware and licensing, and costs more over the life ofthe solution. With lower cost, less network complexity, streamlined deployment andmanagement, and greater out-of-box functionality, Cisco UCS provides a comprehensivemanagement solution to meet your business needs.$1,413,937.88$2,040,418.00$0$500,000$1,000,000$1,500,000$2,000,000$2,500,000Cisco UCS IBM Flex SystemCostinUSdollars80-blade deployment costs
A Principled Technologies test report 15Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemAPPENDIX A – DETAILED COSTSFigures 12 through 14 detail the equipment and costs we used in our analysis. We used configurations thatprovide the maximum bandwidth and high availability features discussed in this paper. All Cisco prices were current as of05/23/2013 from the PC Connection Web site. All prices from the IBM Pre-sales Configurator were also current as of05/23/2013. Standard warranty is included in the list price for both solutions and prices do not include discounts, taxes,or shipping.Cisco product Price QtyTotal CiscosolutioncostTotal IBMsolutioncostQty Price IBM productUCS B200 M3 bladeserver, bare metal$1,495.00 12 $17,940.00 $67,428.00 12 $5,619.00IBM x240 bladeserver, 1 proc, 8GBRAMIntel Xeon processorsE5-2690$2,325.57 24 $55,813.68 $38,748.00 12 $3,229.00Intel Xeon processorsE5-269016GB 1,600MHzDDR3 RAM$310.00 192 $59,520.00 $63,168.00 192 $329.0016GB 1,600MHz DDR3RAM146GB 6Gb SAS 15KRPM SFF HDD$315.55 24 $7,573.20 $8,856.00 24 $369.00146GB 6Gb SAS 15KRPM SFF HDDVIC 1240 modularLOM$719.52 12 $8,634.24 $13,188.00 12 $1,099.00CN4054 10Gb VirtualFabric AdapterVIC 1240 portexpander$288.00 12 $3,456 $9,228.00 12 $769.00CN4054 Virtual FabricAdapter-SW UpgradeUCS 5108 BladeChassis$2,639.56 2 $5,279.12 $1,699.00 1 $1,699.00 IBM Static Server rack2500W Platinum PSUfor UCS 5108$461.27 8 $3,690.16 $5,589.00 1 $5,589.00Flex System EnterpriseBlade ChassisFan module for UCS5108$0.00 8 $0.00 $1,039.00 1 $1,039.00Additional ChassisMgt ModuleServer rack $2,501 .00 1 $2,501.13 $41,798.00 2 $20,899.00Flex System FabricCN4093 10Gb ScalableSwitch2208XP FEX fabricextender modules$4,613.32 4 $18,453.28 $978.00 2 $489.00Redundant FanModuleUCS 6248UP FabricInterconnect + 12pL$15,772.02 2 $31,544.04 $1,956.00 1 $1,956.004 x Chassis 2500WPower ModuleUCS 6200 Series1PORT license$1,387.00 24 $33,288.00 $25,198.00 2 $12,599.00FlexSystem ManagerNode (Redundancy)
A Principled Technologies test report 16Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemCisco product Price QtyTotal CiscosolutioncostTotal IBMsolutioncostQty Price IBM productUCS 6248UP PowerSupply$700.00 4 $2,800.00 $2,240.00 1 $2,240.008FSM Service FabricProvisioning w/3 YrS&S (per chassis)UCS 6248UP ChassisAccessory Kit$150.00 2 $300.00 $12,600.00 1 $12,600.00FSM Standard licenseper chassisUCS 6248UP FanModule included$0.00 4 $0.00 $21,998.00 2 $10,999.00Flex System FabricCN4093 10Gb SwitchUpgrade 2 licenseTotal $ 250,792.85 $315,711.00 TotalCisco cost advantage 20.6%Figure 12: Cost comparison for a 12-blade deployment.8List price from http://www.pcnation.com/web/details.asp?item=QQ0705.
A Principled Technologies test report 19Complexity and cost comparison: Cisco UCS vs. IBM Flex SystemABOUT PRINCIPLED TECHNOLOGIESPrincipled Technologies, Inc.1007 Slater Road, Suite 300Durham, NC, 27703www.principledtechnologies.comWe provide industry-leading technology assessment and fact-basedmarketing services. We bring to every assignment extensive experiencewith and expertise in all aspects of technology testing and analysis, fromresearching new technologies, to developing new methodologies, totesting with existing and new tools.When the assessment is complete, we know how to present the results toa broad range of target audiences. We provide our clients with thematerials they need, from market-focused data to use in their owncollateral to custom sales aids, such as test reports, performanceassessments, and white papers. Every document reflects the results ofour trusted independent analysis.We provide customized services that focus on our clients’ individualrequirements. Whether the technology involves hardware, software, Websites, or services, we offer the experience, expertise, and tools to help ourclients assess how it will fare against its competition, its performance, itsmarket readiness, and its quality and reliability.Our founders, Mark L. Van Name and Bill Catchings, have workedtogether in technology assessment for over 20 years. As journalists, theypublished over a thousand articles on a wide array of technology subjects.They created and led the Ziff-Davis Benchmark Operation, whichdeveloped such industry-standard benchmarks as Ziff Davis Media’sWinstone and WebBench. They founded and led eTesting Labs, and afterthe acquisition of that company by Lionbridge Technologies were thehead and CTO of VeriTest.Principled Technologies is a registered trademark of Principled Technologies, Inc.All other product names are the trademarks of their respective owners.Disclaimer of Warranties; Limitation of Liability:PRINCIPLED TECHNOLOGIES, INC. HAS MADE REASONABLE EFFORTS TO ENSURE THE ACCURACY AND VALIDITY OF ITS TESTING, HOWEVER,PRINCIPLED TECHNOLOGIES, INC. SPECIFICALLY DISCLAIMS ANY WARRANTY, EXPRESSED OR IMPLIED, RELATING TO THE TEST RESULTS ANDANALYSIS, THEIR ACCURACY, COMPLETENESS OR QUALITY, INCLUDING ANY IMPLIED WARRANTY OF FITNESS FOR ANY PARTICULAR PURPOSE.ALL PERSONS OR ENTITIES RELYING ON THE RESULTS OF ANY TESTING DO SO AT THEIR OWN RISK, AND AGREE THAT PRINCIPLEDTECHNOLOGIES, INC., ITS EMPLOYEES AND ITS SUBCONTRACTORS SHALL HAVE NO LIABILITY WHATSOEVER FROM ANY CLAIM OF LOSS ORDAMAGE ON ACCOUNT OF ANY ALLEGED ERROR OR DEFECT IN ANY TESTING PROCEDURE OR RESULT.IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES, INC. BE LIABLE FOR INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES INCONNECTION WITH ITS TESTING, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN NO EVENT SHALL PRINCIPLED TECHNOLOGIES,INC.’S LIABILITY, INCLUDING FOR DIRECT DAMAGES, EXCEED THE AMOUNTS PAID IN CONNECTION WITH PRINCIPLED TECHNOLOGIES, INC.’STESTING. CUSTOMER’S SOLE AND EXCLUSIVE REMEDIES ARE AS SET FORTH HEREIN.