In retrospect, VMware Inc.'s phenomenal IPO may well be viewed as the point at which virtualization reached critical mass. VMware's $27 billion market capitalization indicates that investors anticipate growth far beyond the $1.2 billion in sales forecast for 2007, but the company is not without competition. Citrix Systems Inc. recently spent $500 million to acquire XenSource Inc. Microsoft is busy building its own hypervisor. And smaller virtualization competitors such as Virtual Iron Software Inc., SWsoft and others continue to make strides. So the question is this: Can VMware continue to justify a price for its software that is five to 10 times higher than the competition?The wrong question
Organizations often ask a more tactical version of this question, as in: Which virtualization product is the best choice for deployment? My response is always the same; it's the wrong question. There's an old saying that if you don't know where you're going, you'll likely end up somewhere else. The choice of product should not be the driving concern; rather, the question is whether an organization should virtualize .
By "virtualize," I do not mean setting up a few testing and development virtual machines or virtualizing some old NT 4 servers on their last gasp. I'm talking about the whole shebang: production servers, storage, backup, network, disaster recovery and, increasingly, clients. When an organization embraces virtualization at the holistic enterprise level, amazing things happen.
Take the Mechanics Bank, one of the largest independent banks headquartered in the San Francisco Bay area. A couple of years ago, we virtualized 102 of the bank's 105 servers onto eight VMware ESX Server hosts, and the bank is on track to save $1.6 million in "hard" costs over five years as a result. Mechanics received other benefits as well. The bank now has high availability for all of its servers. IT staff can provision new virtual machines in minutes. The data center space requirements were slashed, a budgeted $60,000 PDU wasn't ordered and one of the data center's two air conditioning units was turned off.
When our consultants pulled out the bank's old servers, they had to put on coats because it got so cold as the power-hungry servers went offline. We have done detailed ROI analyses for scores of organizations considering virtualization at an enterprise level. It's unusual to have a "hard" payback period of more than 24 months, and it's quite common to be well under a year. We've seen finance departments verify payback periods of well under six months.
Virtualization reduces capital expenditures because data centers no longer have to continually replace the majority of servers. It lowers operational expenses by eliminating the need to maintain or administer them. Costs are also reduced in associated areas, such as storage, network, backup and licensing. The power reduction savings alone are often enough to justify virtualization. We even persuaded Pacific Gas & Electric to agree to give monetary incentives to organizations that virtualize, and Nevada Power has agreed to start a similar program with a few of our clients on a trial basis.
Enterprise virtualization quite simply generates exceptional economic savings and other benefits. Delaying its implementation, therefore, is a cause for concern. Every day that servers remain physical means additional risk of server downtime, increased administration requirements and lots more power consumption.
The big virtualization picture
In an organization -- and particularly in a large organization -- decision making can be an irrational process given how human personality plays into the purchasing process. Even though, for example, I may have identified a new car or mountain bike that I want, and even though I may have the money for its purchase, I still am quite likely to futz around for months getting comfortable with my decision before I pull the trigger.
An organization compounds this indecisiveness by requiring consensus from many individuals. Even worse, some of the decision makers may have personal agendas not exactly aligned with the best interests of the organization. One person may be concerned only with his narrow area of IT. Another's primary agenda is to increase her fiefdom. Someone else simply wants to maintain the status quo until he retires in two years. Indeed, I am amazed sometimes that large organizations ever make a decision about anything. Lack of clarity accentuates inertia. The first and most important step toward embracing a virtualization vision is to take a big-picture approach. This involves quantification of both the cost to virtualize the entire enterprise, along with the expected return from undertaking this admittedly significant project."Hard" savings should be analyzed in conjunction with "soft" benefits, such as enabling higher performance and improved security as well as dramatically reducing the time involved in server patching, upgrades, and migrations. I have seen organizations make the decision to virtualize based on the high availability it enables -- let alone all the economic savings.
While some organizations may find it difficult to quantify exact savings and costs, precision is not essential: Savings from virtualization tends to be so great that we are really talking about orders of magnitude. As long as an organization can get reasonably close in terms of anticipated costs and savings, IT management is positioned to make a strategic decision about whether to virtualize and how. Following up the ROI analysis with a well-developed proof of concept validates both the technology and the ROI assumptions and conclusions.VMware's role in the virtualization industry
With a giant head start as the originator of the x86 hypervisor concept, VMware maintains a unique industry position as the dominant player. The company has forged alliances with every major hardware and software manufacturer on the planet, and its products are supported by thousands of consultants. This ubiquitous support translates into a high probability of successful virtualization. It also provides IT departments with the confidence necessary to undertake an enterprise virtualization initiative. On the other hand, even if a lower-priced product might adequately handle an enterprise virtualization deployment in some environments, it clearly entails a higher risk of failure. This in turn necessitates a more cautious and drawn-out deployment approach that is still likely to be less successful than VMware because of a lack of features and universal support. Halting or significantly delaying a virtualization initiative can potentially cost an organization hundreds of thousands or even millions of dollars in unnecessary, ongoing expenses. In this scenario, saving a few thousand dollars on virtualization host servers seems like a pretty foolish gamble. This brings us back to our original question of whether VMware is priced too high. If an organization has determined that it should virtualize only a very narrow environment such as its testing and development servers or a couple of specialty Linux applications, it may be that VMware is not a cost-justifiable solution. If, however, the organization has determined that an enterprise migration to virtual infrastructure is economically advantageous, the cost of VMware appears to be an exceptional bargain. Steve Kaplan is CEO of AccessFlow Inc., a virtualization consulting firm in Sacramento, Calif.. Kaplan previously ran By-the-Bell, an ROI consulting company.