What's the best way to plan for and estimate the ROI of a virtualization project? Find out in this article.

Editor's Note: This article is an excerpt from the book IT Virtualization Best Practices, published by MC Press.

Virtualizing IT resources can lead to savings of over 50 percent in hardware, maintenance, and energy costs in a data center. The transitional cost of virtualizing large-scale IT infrastructures is hard to estimate, however. If not planned correctly, it can often outweigh the benefits of virtualization for several years. This chapter outlines a "divide and conquer" approach to planning and cost estimation for virtualization transformations. This enables an organization to plan and execute complex virtualization transformations while maximizing its return on investment (ROI) within a favorable payback period.

Barriers to Cost-Effective Virtualization

This section offers a bit of background to support the detailed discussions of organizational environments that come later in this chapter. It also summarizes the approach to virtualization we'll lay out in a virtualization pattern.

IT infrastructures in large organizations are highly complex. They tend to have thousands of servers running hundreds of applications on a diverse set of hardware, operating systems, commercial off-the-shelf (COTS) software components, and homegrown applications in a multi-vendor, multi-version environment. Similar applications might have been developed at different times for different parts of the organization, especially when the organization has been created through corporate mergers. Combining similar applications might prove difficult because they are running on different hardware and operating systems. The complexity is further increased by business rules and constraints that are imposed by an organization's environment.

Given that 80 percent of the world's data centers were built before 2001, many of their hardware, operating systems, applications, network, middleware, and data center facilities are outdated or near end of support from their original manufacturers and providers. Many of these components were also designed with earlier business requirements in mind and are not able to scale up or out readily to meet the future computing demands of the business.

Virtualization of the IT infrastructure provides an organization with the opportunity to upgrade and consolidate these components to better address and manage future strategic, tactical, and operational business goals, and IT requirements. The virtualization effort, however, also depends on the particular components and their uses by the organization in a deep and detailed way. For instance, management might forbid a major upgrade to some critical component, such as a payment system, because any disruption could be fatal to the corporation. However, the organization might desperately desire a major upgrade to some other component that has dependencies on the critical system.

Thus, revamping the IT infrastructure and data center facilities via a virtualization transformation can be a very complicated and expensive endeavor. Without careful design, planning, and execution, it can end in unaffordable costs or even disaster.

As a reader of this book, you are probably involved with preparing proposals and budgets for virtualization transformations. You might be used to doing migrations on a small, local basis through direct source-to-target mapping and ground-up cost estimation. These straightforward planning techniques break down during a large virtualization transformation because of large combinations of servers and applications that have interdependencies. In fact, if you stick to traditional techniques, you might not even have time to generate a proposal covering a project of such a large scope. Furthermore, sticking to these traditional estimation approaches deprives you of optimization opportunities in planning and execution by failing to fully realize the economies that present themselves in large-scale transformation programs.

Achieving an optimal transformation in such an environment requires the assessment of a large number of decision variables and transformation parameters and the evaluation of multiple possible paths. We recommend that IT virtualization transformation be managed in a phased approach, with the goal of minimizing risk while maximizing ROI. This requires a meticulous approach to planning and cost estimates, along with comprehensive program management to set up and achieve the results in the plan.

The pattern in this chapter segments the enormous task of virtualization among both resources and time:

• By segmenting the task among your resources—servers, applications, and data stores—you can group your staff's efforts efficiently and apply a single project to multiple resources.
• By segmenting the task in time—among multiple independent projects—you can budget for each project separately and execute the ones with the best chance of success and ROI first.

The scope of each project is determined by examining a number of business and technical variables and objectives. It is managed as an independent end-to-end program, governed by rigorous program management. Each project contributes to the efficiency and progress of the organization as the virtualization transformation continues.

Scenario

To give a sense of the forces affecting a large virtualization project, this section provides a typical client scenario.

The CIO of a financial firm that manages global financial assets is driven by business pressures from her stakeholders to optimize IT infrastructure and data center costs, while continuing to support the company's strategic business goals. These strategic goals are revenue growth, expanding margins, and competitive differentiation via better alignment of business and IT models. The firm has grown over the decades via an aggressive acquisition strategy. This has created several islands of technology in the form of data centers with a heterogeneous mix of hardware, software platforms, and applications.

The firm's IT infrastructure includes over 1,500 applications, hosted on 5,000 servers that are a heterogeneous mix of distributed servers and a handful of old mainframes. Some of the factors weighing down the data centers include the following:

• In addition to the presence of operating systems and hardware from almost every major vendor in the industry, the company's IT portfolio also includes a diverse mix of applications, ranging from COTS applications to those with specific major customizations, as well as those developed in-house.
• The demand at this firm's seven data centers reached maximum power capacity this year. Fifty percent of these data centers cannot be upgraded to increase power density.
• The currently designed data center capacity is not suited for the varied requirements of production servers, versus those of development and test servers.
• The data center landscape has a high degree of fragmentation, with four medium/large locations, three small locations, and many additional "data rooms." This fragmentation also contributes to higher data center personnel costs and facilities management costs.
• The layout of the existing data centers also results in unnecessary operational risks (e.g., dependence on multiple lower-tier data centers), an inefficient disaster recovery model (e.g., requiring additional equipment to replicate data), and a very expensive storage environment (e.g., most of the data resides on expensive tier-1 storage without an effective information lifecycle and data management policy).
• The production infrastructure for individual applications is spread across the multiple data center facilities. The impact of planned shutdowns and maintenance or unplanned outages is high because the staff members don't understand the interdependencies with other facilities well. The dependency of application chains on multiple data centers multiplies the risk of failure and complicates recovery.

Although the IT staff has some projects to address the complexity of maintaining and recovering applications across all data centers, the CIO has recognized that a critical enabler will be the larger job of creating a much simpler data center landscape. Hence, the firm's executive management has given the CIO the charter to consolidate the firm's fragmented islands of technology into three state-of-the-art data centers over a period of two years.

The CIO has decided to leverage virtualization technology to rationalize and optimize IT infrastructure. She has set out to develop a plan for an IT infrastructure transformation with the following strategic objectives (SOs), critical success factors (CSFs), and key performance indicators (KPIs) in mind.

Strategic Objective 1: Reduce Annual Server Operating Expenses

The objective to reduce annual server operating expenses by $8 million and capital expenditures by $40 million will be met using the following CSFs and KPIs:

CSF 1.1: Reduce the server count and associated maintenance/license costs for distributed servers.

• KPI 1.1.1: Reduce the number of physical servers by more than 80 percent.
• KPI 1.1.2: Reduce the number of operating system instances by more than 20 percent.

CSF 1.2: Reduce the IT labor costs needed to support the target state for distributed servers.

• KPI 1.2.1: Reduce server administration staff by more than 20 percent.
• KPI 1.2.2: Reduce the average server burden on administrators by more than 30 percent, through simplification and automation.

Strategic Objective 2: Align IT to Service Level Expectations

The objective to align IT to service level expectations will be met using the following CSF and KPIs:

CSF 2.1: Business platforms will have a business-driven Service-Level Agreement (SLA) and cost breakdown.

• KPI 2.1.1: Service level managers will assign Service-Level Objectives (SLOs) to all servers and publish compliance reports.
• KPI 2.1.2: Cost transparency will be accomplished by applying unit costs to all consumed server resources.

Strategic Objective 3: Simplify the IT Environment

The objective to simplify and standardize the IT environment will be met using the following CSF and KPIs:

CSF 3.1: Leverage repeatable processes, procedures, and technology standards.

• KPI 3.1.1: Post all of the organization's standard technology and services in its IT service catalog.
• KPI 3.1.2: Require explicit approval of all nonstandard technology from the Change Control Board.

Strategic Objective 4: Increase the Agility of IT

The objective to increase the agility of IT to respond to business needs will be met using the following CSF and KPI:

CSF 4.1: Automate and accelerate the operational readiness of new or modified servers.

• KPI 4.1.1: Keep the average request-to-readiness time for new OS images to less than four days.

Strategic Objective 5: Optimize the Use of IT Resources

The objective to optimize the use of IT resources will be met using the following CSF and KPIs:

CSF 5.1: Increase the average utilization of servers.

• KPI 5.1.1: Keep the average CPU utilization across the entire production server landscape at more than 60 percent.
• KPI 5.1.2: Complete provisioning and application readiness of standard offerings in less than five days.

The CIO understands that the virtualization transformation of the server infrastructure will be critical to achieving the strategic objectives. However, she is quite concerned about the cost, complexity, risks, and scheduling of the transformation. Her team has never undertaken a complex IT infrastructure transformation of this nature.