Accelerate server virtualization to lay the foundation for cloud
IBM Global Technology Services January 2012Thought Leadership White PaperAccelerate server virtualizationto lay the foundation for cloudAnalytics combine with standard, repeatable processes for quickertransformation and improved ROI
2 Accelerate server virtualization to lay the foundation for cloudContents This white paper will discuss the role that analytics and defined, repeatable processes can play in improving the virtualization and 2 Introduction consolidation process. It will also list practical steps that organi- 2 Virtualization as a first step on the path to cloud zations can take to virtualize their IT infrastructures—even computing the complex workloads that many companies have avoided virtualizing—and migrate to cloud computing. These steps 6 Moving to cloud computing employ advanced analytical and migration tools developed by 7 Why IBM? IBM Research and deployed in client engagements worldwide.Introduction Virtualization as a first step on the pathIn The Essential CIO: The 2011 IBM Global Chief Information to cloud computingOfficer Study, 60 percent of the more than 3,000 CIOs inter- The benefits of cloud computing are many: it can help accom-viewed said they want to implement cloud computing over the modate end user expectations for technology, overcome thenext five years as a way to grow their organizations, achieve problem of data center obsolescence and reduce IT costs overall.competitive advantage and meet current IT challenges.1 In To lay a sound foundation for cloud computing, organizationspreparation for the move to cloud, companies are initiating must first consolidate and virtualize their server environments.server consolidation and virtualization projects. IBM has found that the cost savings, efficiencies and business advantages gleaned from server optimization can benefit evenTo ease the server consolidation and virtualization process, those organizations not yet ready to consider cloud computing.IBM has developed services-based analytical tools that model In and of itself, server consolidation and virtualization can helpworkloads and outcomes to determine the best way to virtualize organizations extend the lives of their existing data centers,infrastructures. These tools can also help organizations justify increase IT capacity and cut operational expenses.the investment in consolidation and virtualization; and leveragestandards and repeatable processes to accelerate virtualization. Big data promises and challengesMany organizations will need this type of help. A recent IDC The world’s data is increasing by 2.5 quintillion bytes daily,study found that more than 20 percent of the IT managers sur- 80 percent of it unstructured. The information contained inveyed acknowledged lacking the in-house skills required to build Big Data has the potential to considerably enhance customera private cloud. IDC believes that “External data center service and partner relationships, but extracting meaning and trueproviders can help IT with an objective viewpoint, advanced business insight from mountains of raw data is a difficultanalytics (to see what the real issues are versus perceived issues), process—especially when, as often occurs, data is stored inand years of experience in multiple, diverse data center a series of disconnected silos. To accommodate Big Dataenvironments.”2 and integrate it across infrastructures, organizations are continually growing their installed storage. Consolidation, virtualization, cloud computing and cloud archiving can help free organizations from the need to constantly increase their storage capacity.
IBM Global Technology Services 3The question then arises: How do organizations virtualize, con- Modeling workloads in terms of their alignment with overallsolidate and optimize their IT resources? The process typically business goals can help IT professionals correctly size theirentails clearly defining server optimization and consolidation virtualization projects and identify potential problems beforegoals; creating a roadmap for reaching those goals; analyzing and they occur. The modeling effort begins with discovery of serverrationalizing application portfolios; discovering and modeling inventory, configurations, applications, and utilization ratesworkloads to identify which ones are appropriate for virtualiza- to determine which workloads are the best candidates fortion; determining the best ways to virtualize appropriate work- consolidation and virtualization.loads; and building a business case to justify the investment.Implementing a virtualization plan consists of assessment, IBM Fit for Virtualization analytics examine characteristicsdesign, planning, piloting, testing and full-scale deployment of each workload to identify good candidates for virtualization.of those workloads with migration of applications and data to These analytics sort workloads into six categories (see Figure 1)new virtualized infrastructures. based on virtualization complexity. Included in each category is a set of IBM-defined prescriptive actions required to moveThis is a complicated process. But from experience gleaned applications to the virtualized environment.through thousands of client engagements, IBM now believesthat most server optimization efforts can be broken down intothree distinct steps: Workload characteristics●● Discover and model workloads and outcomes to determine Standalone characteristics the best way to virtualize the infrastructure 1 Potential for migration of machines via bulk-move capability●● Justify the investment and validate ROI Basic workloads●● Leverage standards and repeatable processes to define the Potential for bulk-move capability 2 Complexity increases if multiple applications are comingled across server ecosystems optimization process, its methods and goals Complex workloadsDiscover and model workloads and outcomes 3 Due to size, security and risk Managing connections to upstream and downstream applications is a mustIn order to achieve full server consolidation and optimization—including virtualization of the most challenging workloads—it is Middleware remediation and upgrade Multiple applications and servers comprise the workload and all must move together tocritical to understand the current server environment: hardware, 4 reduce test requirements; lack of adherence to standards and versions elongates migrationsoftware, utilization, dependencies, configuration requirements Application remediation and upgradeand business constraints. Only after achieving a thorough Requires application upgrade, with dependence on independent software vendor’s investment 5 in code running on that platform enterprise; resource planning applications typically requireunderstanding of the existing server environment can organiza- upgrades and vendor mandated tools and migration methodologytions develop a truly effective strategy for consolidation and Operating system remediation and upgrade Often requires remediation to resolve operating system-specific callsoptimization. IBM has developed services-based analytical 6 Third-party product usage and availability on target operating system requires operating system upgradetools—including Fit for Virtualization analytics and Fit forPurpose analytics—to help in this work. Figure 1: Fit for Virtualization analytics profile workloads into six scenarios.
4 Accelerate server virtualization to lay the foundation for cloudTypically, organizations find it fairly easy to virtualize 20 or30 percent of their environments—the easiest applications and Fit-for-purpose analytics Workload placementworkloads. However, that leaves up to 80 percent of the server VMwareenvironment running in its traditional physical state. The most Workload IBM AIX® Windows Windows attributes on power® on X86significant opportunity for improved ROI lies in virtualizing on X86 ISV Support 31 18 20those more complex workloads that require the most work tomigrate and that may need remediation and upgrades to operate Performance 40 17 21properly in a virtualized environment. (Virtualization of these Scalability 38 15 18complex applications is also necessary to lay the groundwork SW licenses 39 20 13for cloud computing.)It is a challenging process, but by deploying Fit for Energy 31 19 9 consumptionVirtualization analytics and virtualization methodologies, Overall score 247 170 138IBM has been able to help clients reduce transformation time Five - year TCO US $13M US $31M US $38Mand cost by 30 to 50 percent or more, with a 6 to18 month Five - yearreturn on investment. In addition—and just as important— savings US $25M US $7 US $0virtualizing these complex workloads has helped optimize server Recommend Invest Divest Divestefficiency and resiliency, better aligning the server environmentwith organizational needs and strategies. Figure 2: This is an example of workload placement recommendations usingA second analytical tool, called Fit for Purpose, has been Fit for Purpose analytics. (ISV=Independent Software Vendor.) Individual client results will vary.designed to identify the best-fit platform for each workload.Fit for Purpose analytics assess business requirements againstmore than 30 workload attributes, then recommend the bestplatform for each. (See Figure 2.) This process helps organiza- Optimization reduces server count by 90 percenttions build an effective server strategy and placement plan. A large enterprise client worked with IBM to simplify its server environment, which initially consisted of thousands of serversOften, server consolidation and virtualization activities are taking and more than a dozen platforms. The client hoped to achieveplace within an existing data center. However, the intelligent use a more standardized, flexible and cost-effective serverof analytics can also help when server optimization, consolida- environment. Using the analytics of IBM Server Virtualization Services, IBM analyzed the client’s environment, assessingtion and virtualization are taking place as part of a data center its business requirements against more than 30 workloadrelocation project. Server, application and dependency maps, characteristics. IBM assigned a placement score to eachsuch as those developed by IBM, use analytics to help identify workload, then recommended strategic actions, includingall assets that need to be moved. Information gleaned from divestitures. As a result, the client was able to have plansthese maps aids in moving dependent applications together; to reduce its server count by 90 percent, the number ofin bringing them online together; and in migrating data online platforms in use by 60 percent and operational costs bynearly instantaneously to meet application availability require- 45 percent.ments. All of this helps mitigate risk in the inherently risky datacenter relocation process.
IBM Global Technology Services 5Justify the investment expenditures and returns are provided. The business case is con-Building a business case that quantifies expenditures and tinually refined throughout the consolidation and optimizationvalidates return on investment is critical for justifying funding process to provide the most up-to-date metrics possible.and getting the corporate buy-in needed to initiate serverconsolidation and virtualization projects. There are many ways Leverage standards and repeatable processesto build this case. As seen in the sample analysis depicted in In IBM’s experience, too many organizations fail to clearlyFigure 3, financial analysis tools developed by IBM can provide define their processes, methods and goals for server consolida-companies with a reliable expectation of concrete returns from tion and optimization. The importance of this process cannottheir transformation initiatives. be overstated if these initiatives are to prove successful on their own and lay a sound groundwork for an eventual move to cloud computing. Current Proposed Change Cumulative cost savings over 5 years As discussed earlier, IBM Fit for Virtualization analytics profile Physical numbers of servers 800 272 -66% workloads into six categories. For each category, IBM has a Cumulative operating cost (USD) defined, prescriptive methodology for virtualization. These System management Hardware maintenance $19,546,800 $9,175,169 $11,282,551 $4,962,655 -42% -46% methodologies deploy repeatable processes and automation Facility $4,400,000 $1,900,000 -57% tools developed by IBM Research to aid in virtualization. Power & Cooling $4,185,643 $2,071,893 -51% Cost (US M$) Total operating costs $37,307,612 $20,207,099 -46% One time cost (USD) Hardware purchase/ $0 $605,448 upgrade Transformation $0 $504,010 Simplifying for savings Total one time costs $1,109,458 One time cost + 5 When a large client wanted to simplify its complex, heteroge- Year cumulative costs (USD) $37,307,612 $21,326,557 -43% neous server environment, IBM was able to help. IBM modeled One time cost Power and cooling the client’s server utilization, using best-of-breed tools Five-year savings (USD) $5,981,055 Facility Hardware maintenance and analytics to identify application dependencies, develop Payback period 5 months System management a detailed workload placement plan and model projected infrastructure demands. Additional analytics helped IBM model business, utilization and technical constraints.Figure 3: IBM Anchor financial analysis tools model investment scenarios to Information gleaned from this modeling effort was then loadedfind the optimum server consolidation and virtualization approach to meet the into IBM’s Anchor financial analysis tool to chart total costclient’s financial, business and technical requirements. of ownership and return on investment. The Anchor analysis helped build the business case needed to justify the invest-These tools provide organizations with the information needed ment in server consolidation and optimization. Followingto assist in decision making. For example, these tools can be IBM’s suggestions, the client was able to improve server utilization rates and reduce server count by 80 percent.used to model multiple investment scenarios to find the bestapproach for the client’s specific financial, business and technicalrequirements. Five-year projections of capital and operational
6 Accelerate server virtualization to lay the foundation for cloudThe first optimization process, the IBM Virtualization Fast Path, Moving to cloud computingcan help organizations virtualize many standalone workloads IBM believes that cloud technologies can enable technologicalquickly, in bulk, using rapid migration methodologies. This advances that will help businesses better compete in a constantlyoption is most suitable for organizations needing a quick and evolving global marketplace. Thus far in this paper, we’ve dis-cost-effective way to tackle standalone workloads. It entails har- cussed server consolidation and virtualization. These are thevesting and analyzing information from the organization’s cur- first steps to cloud computing. Next, we will examine the cloudrent environment; performing a Fit for Virtualization analysis; migration process itself.scoping the cost of each element deemed fit for virtualization;determining which workloads will, when virtualized, deliver the By classifying and understanding how specific workload com-highest ROI; and executing a physical-to-virtual transformation plexity impacts migration, and by using a consistent factory-for the technologies best suited to it. based approach (one with defined processes and automation), organizations can drive a more efficient migration to private,The second IBM optimization process, developed for more public and hybrid clouds. To help clients achieve maximum valuecomplex workloads, relies heavily on virtualization to effect a from their cloud strategies, IBM offers a four-step approach.deeper transformation of server environments. This option is (See Figure 4.) This approach is built on IBM technologiestailored to organizations that view server consolidation and that automate the transition to cloud computing.optimization as an initial step on the road toward cloudcomputing. Taking advantage of this option, project leadsfirst work with IBM to frame a virtualized solution for their Private Cloudserver environments—discovering, modeling and assessing Step 1: Step 2: Step 3: Step 4: Inventory Transformation Transformation Implementation/ Clienttechnologies to meet IT requirements, budgets and business veriﬁcation planning design test data centergoals. IBM teams collaborate with the organization’s IT division Client’s environment Rapid migrationto discover and segment hardware, software and workloads; IBM data TCO + analysis Wave strategy center Re-platformdefine IT and business requirements; model and analyze Determine Factory lines workload by Reinstallationinfrastructure demand scenarios; and create a business case migration complexity (repeatable model) IBM SmartCloudto justify investment through comparisons of projected total Analytics + tools + innovationcost of ownership and return on investment. Figure 4: IBM offers a four-step approach for migrating to a cloudAs part of this optimization process, a flexible, virtualized environment.infrastructure is designed to meet current needs and projectedfuture needs. IBM defines a new architecture; deploys a The approach begins with verification and analysis of the client’sstandards-based design methodology to develop and document application portfolio to determine the most effective and expedi-a detailed server design; and develops plans for server construc- ent “waves” of transformation. In the second step, a high-leveltion and wave migration to meet organizational requirements plan is developed to determine migration strategy, migrationand minimize disruptions. Finally, a virtualization solution is methods, target requirements and migration approach. Differentimplemented. Specialized teams will pilot, test and deploy the migration methods are used depending on the degree andvirtualized server infrastructure and then migrate applications complexity of the workloads to be moved to the cloud.and data in accordance with the client’s application availabilityrequirements.
IBM Global Technology Services 7(Opportunities for migration are identified before the comple- In the third step, IBM designs and constructs the transformationtion of detailed wave consolidation plans and schedules, and Fit to a cloud environment. IBM provisions a target test environ-for Virtualization analytics and other unique IBM tools are used ment and then standardizes, virtualizes and modernizes work-throughout the process.) loads using analytics, tools and automation. In the fourth and final step, IBM finalizes cloud migration with remediation andRapid migration is used for the easiest workloads. It migrates testing, migrating workloads to cloud images.operating systems, middleware, codes, applications and data tothe cloud environment as a group, allowing for small adjust- Why IBM?ments as needed. Re-platform is used for workloads of midlevel Virtualization and cloud computing can be complicated pro-migration complexity. It migrates port applications from the cesses. Many organizations realize that to make virtualizationsource operating system to supported operating systems in the and cloud on-boarding as efficient as possible, they need to workcloud environment. This usually requires higher-level code with an outside technology partner. IBM offers the advancedadjustment (for example, UNIX to Linux). Workloads of the analytical tools, deep experience and skills necessary to helphighest complexity are reinstalled. IBM conducts major version organizations achieve the most effective and efficient serverupgrades for operating systems (for example, Windows 2000 to virtualization possible, while concurrently laying the foundationWindows 2008) while migrating for compliance with the cloud for cloud on-boarding.environment. When practical, migration services for cloud drivesthe implementation of each wave as early as possible, and in IBM leverages patented analytics to more quickly and easilytandem with other waves, in order to maximize the delivery virtualize even the most complex workloads—helping clientsschedule and benefits. achieve up to 70 percent virtualization while obtaining ROI in 6 to 18 months. For those clients wishing to move to a cloud environment, we deploy a unique cloud adoption framework and patented workload analysis tool to examine existing environ-Server consolidation and optimization: IBM’s own story ments and determine the right virtualization and cloud comput-At IBM’s strategic outsourcing data center in Lexington, ing model to meet the client’s specific business needs. TheseKentucky, IBM hosts the operations of 20 clients. Examiningthe server environment at this data center, IBM found that capabilities allow us to combine virtualization, standardization60 percent of the 1,500 UNIX servers installed were running at and cloud migration into one event for those workloads that5 percent utilization or less—leaving IBM with 20 times more are cloud candidates. In addition, IBM can provide the rightservers than needed and the expense of operating them hardware and software to help clients fully leverage cloudall. IBM implemented a virtualization program to rectify the computing—whether the organization needs to migrate tosituation and lower operational costs. Through this initiative, an existing infrastructure or first build an infrastructure andIBM was able to virtualize 70 percent of the workloads then migrate.at this data center, increasing server utilization rates to30 to 50 percent of capacity. IBM then conducted an energy- Our deep expertise also brings value to our clients. The experi-efficiency assessment to optimize the cooling environment ence of the IBM virtualization and cloud-computing team helpsand reduce overall energy consumption—driving energy reduce the system risk involved in the cloud-migration process.consumption down an additional 10 percent. As a result of thisinitiative, IBM was able to defer US$50 million in capital costs.