Courtesy: http://searchcio-midmarket.techtarget.com/
The vexing question that periodically intrudes on a midmarket CIO’s agenda — Do I build a new data center or let somebody else take over my diamond-studded ball and chain? — has been brought into sharp relief by the economic recession. With capital scarce and unrelenting pressure to cut costs, data center outsourcing is looking like an increasingly attractive option, market watchers say.
"We are seeing more interest in outsourcing among midsized companies than before and in the $500 million to $1 billion revenue range, a lot more interest," said Rich Matlus, a vice president of research at Stamford, Conn.-based Gartner Inc.
The trend corresponds with other periods of economic decline, when data center outsourcing has increased from about 5% to 7% of Gartner clientele to as high as 10%. But this time, Matlus said, another factor enters the equation: the age of existing data centers relative to current technologies.
"Many were built in the 1970s and 1980s, and they are not the most energy efficient and cannot handle the current processors," he said. "CIOs are left with [the choices of] building a brand new data center, leasing space off somebody else who has a newer facility [or] outsourcing the entire data center."
Indeed, the brainteaser of how to plan and manage a physical facility that depreciates over 25 years for an industry that has seen dramatic change in the past 10 years is making outsourcing in all its permutations — managed, hosted, remote support, colo, the whole shebang — look good to many CIOs, said Andreas M. Antonopoulos, an analyst at The Nemertes Research Group Inc.
Building a data center, especially to Tier 3 or even rarer Tier 4 specifications, is expensive and difficult. Depending on the level of redundancy, costs can range from $2,000 per square foot to tens of thousands of dollars per square foot. Because of the capital investment, the facility has to last a long time to squeeze out the value. And consider the potential for error: A data center built in 1991 to last until 2015 would have required modifications for the Internet in the mid-1990s; in 2000, for blade servers; and in 2005, for what is now called cloud computing.
"I liken it to rocket science," said Antonopoulos, senior vice president and founding partner of Mokena, Ill.-based Nemertes. "It is possible, but only if you have rocket scientists. Failures are few and far between but when they do occur, they tend to be fiery explosions — spectacular in ways you wouldn’t want," such as pricey new data centers that suffer outages or don’t anticipate new developments.
Data center metrics that matter: Cost-effective and efficient
But a data center strategy should not be shaped by just external forces, said Gartner’s Matlus. And that would include the conversation the CEO had last weekend on the golf course with a buddy touting the benefits of data center outsourcing.
"I always warn a client who comes to me about outsourcing to make sure they are doing this for the right reasons," Matlus said.
That list of reasons starts with a sourcing strategy that takes into account what is going on in one’s industry, what kind of initiatives the business has planned and how well one’s current data center strategy can meet those demands, in terms of resources, skills and capital expense.
"If they can do all that, and do it cost-effectively and do it well and are a quality shop, it may not behoove them to outsource the data center," he said, even for companies where the data center is a commodity service rather than a strategic advantage.
Outsourcers run cost-efficient and effective shops, but after they add in their profit, marketing and sales costs, their prices are about average, meaning some shops won’t save a lot of money by outsourcing, Matlus said. "So if you have a shop that is running on below-average costs and at high quality, it does not make sense to outsource."
However, a low-cost, fully depreciated data center that is not performing to users’ satisfaction must be fixed, and outsourcing the whole thing may be a viable option, given capital constraints, Matlus said. Or for a shop that is highly efficient and very high cost, outsourcing could save money.
Comparison shopping: Know thyself first
When a company goes to an outsourcer, it does not have the option of telling the provider how to perform the work but rather what metrics it must perform against. So firms like Gartner and Nemertes typically counsel companies that the service-level agreements (SLAs) for uptime and availability from an outside provider should meet and exceed the SLAs for the business — and be enforceable. The contract should lay out objectives of work, delineating which roles IT will maintain and which will be performed by the outsourcer, and so on.
The problem with making that cost and performance calculation — or negotiating an effective contract — is IT shops often don’t have a good grasp of how long it takes even their own people to perform certain tasks, and so can’t make a fair comparison, said Bill Peldzus, vice president, data center and BC/DR at Glasshouse Technologies Inc., a Framingham, Mass.-based consulting firm. Many IT shops don’t even know what’s in their current service-level agreements.
"They have no idea, on something as interesting as, ‘I need 100 more gigabytes of storage and how long will it take to get that,’" Peldzus said.
And it is not just the small shops without the sophisticated business metrics to measure cost and efficiency that are in the dark. "Let’s just say I have seen customers on the Fortune 10 list that do not know how long it takes to provision storage," Peldzus said.
Gartner’s Matlus agreed, telling the story of a recent Fortune 100 client whose data center performed at a high level but was also a money pit, so it was considering outsourcing. A benchmark study uncovered that though the data center employed tape robotics to automate storage, it had kept all its tape hangers, basically obliterating the cost savings reaped from automating.
Mainframes, brain drain, baby boomers, Linux, virtualization
In the next few weeks, SearchCIO-Midmarket.com will explore why and how midmarket CIOs are outsourcing all or parts of their data center, and will talk with CIOs who are equally passionate about keeping the data center under tight internal control.
Meantime, industry experts say there are plenty of reasons other than a cash-strapped economy driving midmarket companies to look outside their four walls for data center support. Companies that are having trouble finding people to support their mainframes have been calling Gartner for advice, Matlus said. Young people weaned on Linux, Unix and Wintel don’t want to learn mainframe systems, and as the economy has worsened and companies have been forced to lay off employees, people 45 and older have been disproportionately hit, exacerbating the shortage of mainframe support.
Conversely, companies trying something new — like Linux or virtualization — that don’t have the internal skills are going for outside help, rather than hiring. In addition, virtualization is both extending the life of internal data centers and paving the way for outsourcing to another provider, as companies get used to virtual computing environments.
And companies with a single location are looking into obtaining redundancy by moving the data center off-site, whereas companies with multiple locations can realize economies by moving to an off-site data center, assuming they have sufficient bandwidth.
Yet many stop short of utility or cloud computing for their core systems, as security remains a concern, even for early midmarket embracers.
The CIO of a fast-growing online services business has unloaded his corporate email to the Microsoft cloud, for instance, and is looking to put financial systems in the cloud. And additional cloud storage on the cheap for nonsensitive data would be great. But anything with customer information? "At this point security is so paramount to our customers, I wouldn’t entrust that to anyone else," he said.