Facilitating data center virtualization success with an ROI analysis

Several years ago, Pittsburgh health care facilities were trying to fight the rapid spread of Methicillin-Resistant Staphylococcus Aureus (MRSA), a deadly hospital-acquired infection. Dr. Richard Shannon, Chief of Medicine at Allegheny General Hospital, needed to purchase 100 patches at a cost of $5 each to pilot a method for slashing catheter induced infections, but couldn't get budget approval. It was only when he presented an economic analysis for hospital administrators showing that they were losing $17,000 per infection that Dr. Shannon received $500 for the patches – along with an additional $1.3 million.

Demonstrating abundant cost savings is effective in both health care and in IT. An in-depth ROI analysis helps secure the funds required for a strategic approach to data center virtualization while also assisting the virtualization champions successfully navigate the organizational resistance and politics that accompany significant change.

Why do an ROI Analysis when the Economic Advantages are Obvious?

Combining VMware vSphere with Intel's Nehalem CPUs can commonly accommodate virtualizing 100 physical servers onto six 2-CPU/96 GB hosts – including one for redundancy. Although obvious that huge savings result from 94 less servers to power, cool, maintain and periodically upgrade, an ROI analysis should always be conducted:

It encourages a strategic approach to virtualizing the data center, enabling a much more elegant architectural design. Traditional silos of compute, network and storage are optimized while minimizing the point solutions to manage.
It promotes virtualization as both a unifying platform for disparate data center technologies as well as the foundation for more effective disaster recovery, desktop computing and cloud computing services.
It assists in determining the feasibility of implementing a comprehensive enterprise virtualization project from the start. Regardless of deployment timeline, expenditures incompatible with with an optimized virtual infrastructure can be minimized ranging from servers to air conditioners to certain types of storage and networking equipment.
It reveals both costs and opportunities that might otherwise be overlooked. A health care organization, for example, capitalized on Microsoft's advantageous virtualization licensing policies to enable savings of nearly $1M in its Enterprise Agreement negotiations.
It provides a baseline against which the IT organization can measure the financial aspects of its virtualization success, helping to build credibility.

How Does a Data Center Virtualization (DCV) ROI Analysis Work?

ROI is the return an organization can expect to achieve on its investment over the analysis period. Expected costs are identified for operating the data center over a given number of years under two scenarios: physical and virtual. The virtualization investment is subtracted from the delta in total cost and then divided by the investment.

Figure 1 shows a savings calculation example for virtualizing 100 physical servers onto eight 2-CPU VMware vSphere hosts. It assumes a 5-year analysis period along with server growth of 10 new machines annually. Given the tough economic climate, physical servers are refreshed only once every 5 years rather than the manufacturer recommended 3 years. Maintenance contracts, following 3-year manufacturer warranty expiration, are utilized for ½ the servers at an annual cost of $1,000 each. The vSphere hosts are refreshed every three years and the new models will accommodate far more VMs than today's units. The physical servers cost $10,000 each including tax, shipping, rack space, cabling, power whips, SFPs, network switch ports, SAN switch ports, HBAs and generator and UPS slices. The vSphere hosts costs $17,000 each. A $65K budgeted PDU is no longer required.

Expected costs over the next five years on the physical side come to $2.7M, while the virtualized data center requires only $440K, including the cost of virtualization software maintenance. This leaves projected savings (or more accurately, cost avoidance) of almost $2.3M.

Figure 1: 5-year savings/cost avoidance from virtualizing 100 servers

Figure 2 shows a sample virtualization investment including first year software maintenance fees along with storage and implementation/ training costs. The total expected investment comes to $480K vs. savings of $2.3M. The 5-year ROI is therefore ($2,270,036 - $477,018)/$477,018 = 393%. The payback period of 12.2 months measures how long it takes for the additional savings to cover the investment.

Figure 2: Investment Required

This simple ROI analysis doesn't factor in variables such as the time value of money, but it also doesn't consider the myriad qualitative benefits of virtualization such as high availability for all servers and potential elimination of clustering requirements; reduced staff time for tasks such as provisioning servers, patching and hardware trouble-shooting; and the potential for improved performance, security and automation. It also excludes potential benefits from incorporating disaster recovery, desktops and cloud computing services.

One of the luxuries of a data center virtualization ROI analysis is that it is generally unnecessary to quantify more than the obvious "hard" savings – at least in terms of justifying the initiative. Nonetheless, it is typically worthwhile identifying the many benefits, as well as new challenges, that virtualizing a data center entails as heightened clarity can only increase the level of success.

What about TCO (Total Cost of Ownership)?

TCO, popularized by Gartner in the late 1980s, is used to estimate the cost of a product over its lifecycle factoring in both direct and indirect acquisition and operating costs, although not savings. TCO is particularly applicable for comparing products that provide similar functionality such as network switches or thin-client terminals. The effectiveness of TCO rapidly diminishes when attempting to evaluate products pivotal to the success of alternative technology platforms.

An organization's approach to virtualization can determine which metric to use. TCO is a valid calculation, for example when comparing Cisco's UCS as simply an alternative to traditional servers. But if its faster virtual infrastructure provisioning and enhanced VI management capabilities are perceived as essential to the virtualization value proposition, then an ROI analysis – in this case quantifying the benefits, is more applicable. TCO is often a good comparison for individual products if deploying virtualization as a point solution. Approaching virtualization as an enterprise platform, though, makes an ROI analysis the only reasonable means of evaluating essential components of that solution.