Transcript of "Data center flexibility and efficiency: increasing the business value of IT"
IBM Global Technology Services February 2012Thought Leadership White PaperData center flexibility and efficiency:increasing the business value of ITIncreasing demand for services provides opportunities to simplify operations
2 Data center flexibility and efficiency: increasing the business value of IT Contents been reduced from more than 15,000 to 4,700. The result? From 2006 to 2011, IT transformation and optimization efforts 2 Introduction yielded US$1.25 billion in savings. 2 Building consensus: asking the right questions for the business Global data center integration has also allowed IBM to improve operational efficiencies by increasing security and resilience, 3 Crafting the strategy: discovery and analysis mitigating risk and delivering applications and information to 4 Turning plan into reality: continuous transformation employees in ways that increase productivity and collaboration, execution while providing the building blocks for further innovation. 9 The results: savings and improvements Of course, no business environment remains static. Continual 10 What comes next: plans for the future refinement of the global data center strategy is needed to 10 Applying the lessons respond successfully to changing business needs that range from regulatory requirements to new acquisitions to enhanced 11 End rewards for the CIO customer relations. 11 For more information This white paper outlines the steps IBM has taken to update our data center strategy, build a unified plan and execute againstIntroduction that plan. This paper also examines the end results, plans forCost reduction. It allows organizations to survive tough years the future, and best practices uncovered along the way. Forand increase profits during good years. It provides an opportu- IT executives who are looking to re-engineer their strategiesnity to innovate, as well—becoming the vehicle through which and processes to create tighter global integration with businessIT executives discover new ways to restructure for growth to needs, the information in this white paper can be leveraged torespond to changing business requirements more quickly, help create an integrated data center strategy and plan, reduceimproving service quality along the way. costs and improve enterprise-wide ability to adapt to change.Cost reduction was the original impetus behind IBM’s efforts to Building consensus: asking the rightintegrate our internal data centers around the globe. The pro- questions for the businesscess started in 1997, when IBM created a global office of the In 2009, an IBM Academy of Technology study made it clearCIO, transferring governance from a geography-centric model that IBM needed to continue to improve the alignment of tacti-to a centralized one that encompasses all IT services for all cal data center initiatives with strategic goals in order to opti-employees. In the ten years from 1997 to 2007, IBM has seen mize business performance. Definitions of what comprised asignificant improvements in operational costs and efficiencies strategic data center, the business logic used in selecting theas a result. Hosting data centers have been consolidated from data centers, as well as the required services and capabilities,155 to five strategic sites, 31 networks have been combined into needed to be documented to allow IBM to achieve more opera-one globally managed network and deployed applications have tional flexibility.
IBM Global Technology Services 3With more than 400,000 employees accessing data and applica- own data center strategy may be different, but will still helptions that were known to be running outside of strategic data you build tight linkages between business strategies andcenters due to acquisitions, organic growth and skunk works technical execution.projects all over the globe, it was important to choose a manage-able starting place to update IBM’s data center strategy. Decision model tenets ●● All data centers should be governed under a single, globallyIn order to help overcome stakeholder’s assertions that data cen- integrated technology delivery modelters needed to be located within each geographic region, the ●● To assure the highest value to the business, total cost ofoffice of the CIO hired “outside” data center strategy experts ownership (TCO) should be measured at a global corporatefrom IBM Site and Facilities Services. This team works with level, not by individual workload, application or data centerclients to use objective, fact-based tools and analytics to guide ●● Appropriate business continuity and security measures shouldexecutive agreement on the decision model that is used to make be applied across all data centers—matching measures to thestrategy recommendations. This allowed the office of the CIO criticality of data and applications within each serverto engage key owners during the analysis process—increasing ●● Location of workloads should be determined by overallstakeholders’ confidence that their individual needs were being business objectives and data center efficiency, not by the usermet in a way that provided optimal value to IBM as a whole. or asset “owner.”The Site and Facilities Services experts reached across all of the Crafting the strategy: discovery andgroups involved to solicit input and discuss over 300 potential analysisdecision criteria, arriving at a universally agreed upon set of key Taking inventorycriteria that provided the best direction for the business. For One of the tactics recommended as part of the globally inte-IBM, these items focus on aspects of responsibility, financial grated governance model was the development and execution ofrequirements, workload and governance and provide clear and a process for maintaining a current and accurate inventory ofoverarching executive guidance that enables the technical teams servers, data and applications—and their locations. This pro-to execute with a high level of confidence that the results will vided the input needed to determine which data centers to holdlink positively to IBM’s business strategy. As each organization or vacate. Priority was given to establishing an accurate and up-has different business strategies, the criteria developed for your to-date inventory of every asset in use within the specified scope for North America and Europe before expanding to include Asia Pacific and Latin America.
4 Data center flexibility and efficiency: increasing the business value of ITAn accurate global inventory was harder to achieve than antici- data center integration strategy and associated processes thatpated as inventory databases did not always contain applications would achieve the goal of enhanced flexibility through optimiza-that were created and maintained outside the office of the CIO, tion. Important strategic elements included:and did not count servers which had been provisioned by indi-vidual business units—often outside of raised-floor space in the ●● Site selection for target migrations based on overallidentified data center sites. IBM business needs. Three strategic, global data centers would support 80 percent of the application portfolio. TheThe Site and Facilities Services experts applied the time-tested remaining portfolio, including unique workloads to supportrule that says that 80 percent of an organization’s applications manufacturing and regional requirements for security andare generally delivered via 20 percent of the data center capacity. privacy, would be housed in special-purpose data centersThey started with the largest locations where internal IBM data located to match the assets housed.and applications were located, and then worked out until they ●● Growth in the strategic data centers would be optimized byhad covered all the locations in scope. virtualizing servers and recovering existing space consumed by aged and/or redundant servers.Whenever possible, the identification of applications and infra- ●● Existing data centers would be vacated as indicated to mitigatestructure was automated using the enterprise portfolio registry, risk of critical applications being housed in less resilient orserver security database (mixed address database) or the applica- secure locations and consolidating expenses to the strategiction inventory process and repository utilizing Tivoli® data centers.Application Dependency Discovery Manager (TADDM) and ●● A continually updated service catalog would be established toCognos® business intelligence reporting. Purpose-built analysis help assure end-to-end IT service management, allowing thetools developed by IBM Research helped speed up the process, right level of hosting support and services to be applied to allalong with helping to assure that all servers, applications and assets.data were properly examined and accounted for. Informationcaptured included the server count by type, data and applications Turning plan into reality: continuouson each server and whether each server platform was the most transformation executionefficient one for the associated assets. Information on server real With the strategy and decision model tenets established, theestate (in square feet), energy usage and network connectivity office of the CIO executed the processes as shown in Figure 1 towas also captured. simplify the data center infrastructure and manage the associated services. This was accomplished by focusing on four major areas:Building the data center strategy migration and consolidation, storage growth, flexible deliveryOnce accurate and up-to-date inventory information was avail- and end-to-end service management.able, this information was filtered through the key decisionmodel tenets (see the “Building consensus” section) to create a
IBM Global Technology Services 5 In IBM’s data center strategy, technology and business requirements work hand-in-hand and are continually reviewed for flexibility and efficiency Build consensus Balance business requirements • Enable business growth • Ensure application availability • Improve operational efﬁciency Craft data center strategy Decision model tenets • One globally integrated technology delivery model • TCO measured at global level • Continuity and security applied to each data center, matching measures to data/application criticality • Location of workloads determined by business objectives and data center efﬁciency Execute per technology objectives Continuous transformation Manage the services • Create end-to-end IT service management with standard services Simplify the infrastructure • Migrate and consolidate servers • Manage storage growth Achieve ﬂexible service delivery • Server type matched to data/application needs — x86, UNIX, mainframe • Cloud delivery where applicableFigure 1: By applying consistent governance and processes through the data center consolidation project—and continually closing the loop by evaluating resultsagainst the decision model and strategy—IBM has been able to fully leverage the opportunities presented by optimization.
6 Data center flexibility and efficiency: increasing the business value of ITMigrate and consolidate to best-fit platform “Fit for purpose” were the bywords when making applicationOne of the most significant ways to minimize capital expendi- consolidation decisions. Is the application currently being usedtures and measure total cost of ownership (TCO) savings at a to access and work with data the best fit from both a user andcorporate level was to migrate and consolidate workloads to the corporate perspective? Is a newer, more effective version of thebest-fit platform. Specific workload recommendations from the application already available within the organization? Is there astrategy team were used during execution, including: standardized application that can do as good a job, or better?●● To deliver optimal efficiency and cost savings, virtualization Based on the fit-for-purpose evaluation, decisions were made to should be the standard keep or retire existing applications, with retained applications●● Utilization of each data center should be maximized to avoid being made more widely available and consistent through virtu- unnecessary expenditures or new capacity alization and relocation into one of the strategic data centers.●● Automation and processing tools should be used to minimize conversion costs While consolidating workloads takes hard work and expertise,●● Optimal TCO can be achieved by eliminating redundant, IBM has found that the results are well worth the effort. Since overlapping or similar IT services. the project’s beginning, scope has increased from 3,900 servers to over 15,000 with significant advances in efficiency. ResultsWith the data center inventory in hand, the execution team was through the end of 2011 are as follows:able to eliminate significant up-front effort by contacting appli-cation owners to sunset applications and decommission servers ●● Over 6,500 servers have been consolidated and virtualized,that were no longer in use. Initially, over 500 servers in 377 cities bringing the total number of servers virtualized to 65 percentwere found in the wild—belonging to an individual location or of the total inventory. When the 2015 roadmap is completed,business unit and operating outside strategic data centers. As the 90 percent of the total servers will be virtualized.team expanded their focus to include Latin America and Asia ●● TCO has been reduced by a dramatic 70 percent—achievedPacific, they found that nearly 30 percent of the application by increasing average utilization from 10 to 60 percent andportfolio and equipment could be retired—avoiding the costs resulting in approximately US$50 million in total costassociated with migration and virtualization. The remaining take-out and US$100 million in savings.servers and applications are to be migrated to the strategic data ●● Contribution to IBM’s environmental savings has beencenters that will support 80 percent of the application portfolio enhanced, with over 30,000 megawatts of energy saved peror to the special-purpose data centers. year and 74,000 square feet of floor space returned to the business.The team knew from prior consolidation efforts that a work- ●● Enhanced migration efficiency with the average migrationload-centric view of the data center would prioritize the low- time being improved by 55 percent—taking 90 instead ofhanging fruit, but would fail to optimize floor space and leave a 200 days—along with a 72 percent reduction in migrationSwiss-cheese-like pattern of missed servers. Instead, the team costs per server.used a location-centric view of the migrations—literally usingyellow caution tape to delineate the servers in scope within anindividual data center. This enabled the elimination of datacenter fixed infrastructure to drive further benefits.
IBM Global Technology Services 7Manage the explosive growth in storage foundation to leverage cloud technologies for rapid and f lexibleWith over 10 petabytes of storage in production use and deployment of resources and accelerated and efficient service15 petabytes allocated for backup, managing the projected delivery. For IBM’s internal data centers, the adoption of cloud25 percent compound growth in storage at flat cost is an integral is shaped by workload affinity and characteristics and deployedpart of achieving the optimization and TCO objective. Smarter on a best-fit basis across all delivery models and platforms.information lifecycle management (ILM) techniques and essen-tial storage technologies are a key part of IBM’s execution Virtualized private cloud has been selected for workloads thattactics to: are suitable for virtualization, but require a more robust level of security and control than can be supplied in a public cloud envi-●● Reduce volume of data to be stored via advanced techniques ronment. Examples include workloads that depend on sensitive for compression and de-duplication data normally restricted to the enterprise, workloads composed●● Maximize utilization of available storage through virtualization of multiple codependent services, and those requiring a higher and thin provisioning level of auditability and accountability. Other candidates for vir-●● Achieve the right balance between efficiency of data access tualized private cloud include workloads based on third-party and storage cost through automated tiering and migration. software which does not have a virtualization or cloud-aware licensing strategy, those that require detailed chargeback orApplying storage tiering to identify the correct data placement— utilization measurement and workloads requiring flexibilityand managing the migration using automation and tiering tools and customization.developed by IBM Research—IBM has been able to achieveorders-of-magnitude improvement in storage efficiency. For Virtualized public cloud has been chosen for workloads such asexample, migrating 57 terabytes of data now takes just six hours, web infrastructure applications (wikis and blogs), email and col-instead of the previous 235 hours—a 98 percent reduction in laboration workloads, development and test workloads, batchtime and effort. processing jobs with limited security requirements and Software as a Service (SaaS) applications. Other virtualized public cloudFlexible service delivery with cloud candidates include high-performance computing workloads withThe workload recommendations, centered on flexibility of loca- low security requirements, storage-intensive workloads with lowtion, directed that applications be coded to avoid tying them to a security requirements, and secure offsite backup/restore solu-specific infrastructure location or platform. Along with making tions for data protection and disaster recovery.virtualization a standard for implementation, this sets up the
8 Data center flexibility and efficiency: increasing the business value of ITThe benefits of deploying cloud within IBM’s internal data ●● Management of all data centers should utilize a consistent setcenters have been significant, including: of tools to allow for maintenance and support efficiencies. ●● Equitable billings for service should be assured to encourage●● For development and testing—the time to set up an image has users to seek more effective ways to store and manage their been reduced from five people-days to just one. As a result, data and applications. over 100 new server images per month are provisioned. Nearly 100 percent of the requests for server provisioning/ The establishment of a single point of governance within the de-provisioning are coming through the development and office of the CIO over a decade ago paved the way for imple- test cloud as opposed to the previous manual-build request mentation of enterprise-wide IT service management (ITSM) process. This improves the speed, flexibility and level of for internal workloads. ITSM shifts the focus from one-off self-service. Complementing this is a 50 percent reduction builds and stand-alone applications to standardized services in IT labor cost and a 75 percent improvement in capital and end-to-end integration of the portfolio. Rather than a focus utilization. on service level agreements (SLAs) for individual applications●● For the desktop—tailored thin clients (customized by role) are and siloed cost management, ITSM provides for the use of key being deployed to 2,000 users in our China development labs business process service level objectives (SLOs) and the ability running on a private desktop cloud, with plans to expand to an to measure—and reduce—total cost of ownership organization- additional 2,000 users for an estimated 20 to 30 percent wide. As a result, top priority can now be put on relevance to savings per user. the business, rather than the performance of individual parts●● For storage—automated self-service provisioning, flexible of the infrastructure. The goal is to support 80 percent of the pricing, elastic capacity and advanced virtualization supports application portfolio with standard services in a single catalog. over 130,000 users accessing email, collaboration and other applications. No longer are all applications treated the same and given the same level of support. Instead, they are classified by criticalityEnd-to-end IT service management and level of support needed, allowing a corresponding matchFour workload recommendations drove the improvements in the in service. Application owners select from one of four levels ofend-to-end IT service management process. support—or tiers—based on the required level of availability, support, redundancy and monitoring needed. Over 80 percent●● Location of workloads should be determined by overall of the applications have been classified in tiers 3 and 4, allowing business objectives and data center efficiency, not by the user service and support cost to be dramatically reduced. Centralized or asset “owner.” service management of the application portfolio has also resulted●● Applications should be provided with disaster recovery in a 36 percent improvement in average deployment cycle time solutions that leverage common infrastructure, tools and for simple applications so far. methodologies.
IBM Global Technology Services 9The results: savings and improvementsExecuting the enhanced strategy for internal data centers has IBM now governs internal data centers under a single, globally integrated technology modelallowed IBM to continue to reduce costs. More importantly it 3 global data centers now support 80% of applicationshas paved the way for operational efficiencies that provide newopportunities for innovation and operational flexibility to meetour changing business needs. Thus far, the effort has yielded thefollowing improvements (also shown in Figure 2):●● Agreement on the number and location of strategic data centers—optimizing the cost and performance of the applica- tion portfolio to the best interests of the corporation as a whole. This has required greater visibility into the server inventory for production workload, allowing for the mitiga- tion of risk for servers not on a hardened floor in more than 290 cities. Continual transformation to deliver optimization through ﬂexibility●● Migration and consolidation of applications to the best-fit server platform, reducing the total cost of IT operations by Reduce cost of IT operations Enable rapid and ﬂexible deployment through total cost of ownership of resources via cloud—65% of 70 percent. This includes returning 74,000 square feet of (TCO) reduction application portfolio eligible for self-managed cloud raised floor space from internal use to revenue-generating use Manage 25% compound growth in Support 80% of the application data at ﬂat cost portfolio with standard services for strategic outsourcing clients, deferring over US$70 million in capital costs to build additional data center capacity.●● Managing growth in storage volumes to support 25 percent annual compound growth at a flat cost. Figure 2: Consolidating and optimizing its internal data centers has allowed●● Flexible service delivery with rapid deployment of resources IBM to realize significant savings across all areas of data center implementa- tion and maintenance within the cloud in hours versus days, and the use of cloud to implement lower tier applications.●● Creation of end-to-end IT service management to support IBM has also realized benefits beyond the numbers, such as 80 percent of the application portfolio with standard services. improved continuity and resiliency, better mitigation of risk and a more thorough and accurate asset inventory.
10 Data center flexibility and efficiency: increasing the business value of ITWhat comes next: plans for the futureThe pace of change will continue into the foreseeable future 2. Find your assetswith continued need to improve IT operations and respond to It can be extremely difficult to identify and capturethe changes in the business requiring IT support, such as organic complete inventory information for all data centers andgrowth, increasing acquisitions, new products and new markets. assets. Accuracy and consistency may be the responsibilityWith results such as those outlined in the previous section, of many people, or no one at all. The best and easiest wayIBM’s roadmap through 2015 will continue to build flexibility to create and maintain an accurate and current inventory isby leveraging optimization, focusing on three key areas: to use analysis and automation tools. 3. Create a single governance focal point●● Optimization of global hosting into the strategic data centers IBM’s consolidation efforts only became possible after that have been identified. This includes completing the the selection of a single point of governance (the office of migration of ~3,000 servers, exploiting the latest server the CIO). When decisions about location and technology technologies and continuing with virtualization on new servers requirements were shifted from application developers and as a standard practice. Implementation of storage tiering and business units to the office of the CIO, consistent processes, deployment of IBM DB2 Analytics accelerator for faster tools and techniques could be applied—resulting in far business insights will also be continued. greater efficiencies and cost savings than would otherwise●● Fully leveraging new asset consumption and delivery models have been possible. by expanding usage of cloud technologies—65 percent of the 4. Keep your eye on floor space already consolidated application portfolio has been determined Traditional metrics for transformation programs do not to be eligible for self-managed cloud. always ensure optimization of data centers and assets.●● Increase in development and deployment of applications Along with improvement measurements for items such as through standard services, as defined by the new end-to-end availability and security, all transformation efforts should IT service management model. include reclamation of data center floor space as an indication of success.Applying the lessons 5. Design for application and workload mobility Mobility of workloads is essential for the way work getsThe learning curve: management strategies to consider done today. Without it, high availability, business continuity and technology currency are all difficult to achieve. At IBM,1. Think and act globally all future work and investments are being designed and Within many global corporations, an assertion exists that evaluated with application mobility as a key requirement to be a true geographic unit, you need a data center within to avoid coding applications in a way that ties them to a your unit boundaries. By separating the selection of specific infrastructure location or platform. strategic data centers from the organization’s financial structure and making decisions based on a consensus on financial feasibility, business capabilities and physical/ technical capabilities—and, most importantly, involving all the players in the planning process—IBM was able to not only overcome resistance, but transform resistors into enthusiastic participants.
IBM Global Technology Services 11Just as interesting as the best practices are the lessons learned capabilities to help,1 it comes as no surprise that many CIOsduring this period, and how to apply them as you plan and view consolidation of their technology infrastructure andexecute your own data center consolidation. organization as the future model for data center operations. Centralized infrastructures and processes enable optimizationA migration strategy of going after low-hanging workload fruit of shared services that, in turn, provides economies of scale tois only as good as the fruit itself. Initially, IBM considered an cut costs and increase return on IT investment.application-focused view of our data centers and other internalservers to select candidates for virtualization and migration. But More and more, CIOs’ level of success and influence withinthis approach did not lead to the desired reclamation of floor their organizations is being measured by how well they managespace (see learning curve highlight #4) or reduction in TCO. their data centers to raise the return on IT infrastructure invest-By executing consolidation efforts using a location-led view, ment, expand the business impact of the data centers and makeIBM was able to get a better understanding of all our internal innovation opportunities real. IBM hopes that this white paper,data center assets, the dependencies between the assets and the based on a real-world execution of a data center consolidationopportunities for optimization—resulting in significant floor strategy, provides a framework that CIOs and other IT execu-space reclamation and a healthy reduction in TCO. tives can leverage to deliver cost efficiencies, savings and opportunities for breakthrough thinking to your ownIncluding the business case in the decision process increased organization’s efforts.the options and the opportunities. Using the business case, datacenter capacity that had been used for revenue-generating For more informationworkload—but had capabilities that made it better-suited for If you would like a more detailed exploration of IBM’s internalinternal workload—could be reallocated, resulting in a more data center story, please contact your IBM marketing representa-efficient and cost-effective use of resources for all parts of tive or IBM Business Partner to schedule an in-person presenta-the business. tion by one of our experts. You can also request information on the IBM offerings, services and tools designed to makeVirtualization is good, but not having to virtualize is even data center consolidation more efficient and effective.better. Making the first decision point for every application aneither/or—keep or retire—IBM was able to sunset a healthy per- Learn more about these offerings by visiting the followingcentage of its application portfolio, avoiding the costs associated websites:with migration. ●● ibm.com/services/smarterdatacenter ●● ibm.com/services/siteandfacilitiesEnd rewards for the CIOWith over 50 percent of CIOs surveyed in the 2011 IBM GlobalCIO Study indicating that they plan to participate in reengineer-ing their business, and are looking for new expertise and