Deciding to outsource

IT executives tell how they've made their choices.

When Karl Kaiser became CIO of Minneapolis five years ago, he found the IT infrastructure was not as capable of supporting the business of running a city as he - and the city constituency - wanted it to be.

"Sixty percent of my budget and management energy went into just keeping the infrastructure alive," says Kaiser, noting that his staff was overwhelmed with break/fix duties inherent in the city's multivendor environment and seemed more concerned with gee-whiz technology than serving the city.

Minneapolis' 4,000 city workers, including police officers, firefighters and government officials, needed more than that. "They were looking for services that went well beyond installing and fixing computers. They wanted more applications, especially Web applications," he says.

Besides new Web initiatives, Kaiser had to fund a disaster-recovery program, find a new space for his data center (which was ending its lease with the county) and staff the city for 24/7 support - all within a tight budget. Kaiser took a hard look at outsourcing and decided it made sense - to a point. He sold his entire desktop, server and network infrastructure - and its management - to Unisys, while keeping application development in-house. The result, he says, is a happier constituency and an estimated $20 million savings over the seven-year life of the contract.

"I decided to get out of that business because the information technology assets and the associated support functions in my mind are a utility. It's like you come to the office and switch on the lights. The light comes on, but that doesn't mean you need to own the power plant," he says.

Minneapolis is not alone. Faced with similar prospects and the need to move quickly to next-generation data center environments, many organizations see the benefits of offloading mundane operations tasks while focusing on the value IT brings to the business.

Starwood Hotels and Resorts Worldwide, for example, outsourced its hardware and network infrastructure to HP in an effort to ease its move off a mainframe and onto a next-generation computing environment that features Web services running on Unix and Linux servers (see related story ).By offloading the operations side of the house, IT could focus on developing core profit-generating reservations and loyalty systems applications, says Tom Conophy, CTO at the White Plains, N.Y., hotelier.

"We wanted to make sure we kept the thought leadership within Starwood," Conophy says. "You can't just go out and buy our reservation system, like you would any [point of sale] or CRM application. I would never outsource a custom-built application that is extremely crucial to our business."

Because of the critical nature of the applications, Conophy has retained ownership of the hardware on which they run. He has, however, offloaded hardware support. "We still look at the Unix and Linux configurations ourselves, and then work with HP to get those engineered, configured and established. We're not quite at the point where we're willing to toss HP the keys - the applications are too custom and too crucial, and not something HP could get certified in."

Kaiser's view is more extreme. Because Unisys now owns the city's computing resources, it has the onus of gaining efficiencies for the customized and regulated municipal government environment. "Unisys bought my 150 servers and put them in its data center, and it has a grand plan for consolidation and virtualization," Kaiser says. "But I don't really care. It could carry my data with Nike sneakers from one corner to another as long as it meets my service levels."

Apple to apples

Deciding whether to outsource your next-generation IT infrastructure goes well beyond studying criticality, as companies typically have done when considering what to outsource, experts say. The new litmus test is whether IT has become a service operation. Only when an organization views IT as a service can it truly measure the costs and weigh the value of outsourcing, they say.

If an organization views IT as a shared resource, then business units no longer expect or need their own application servers, explains Andreas Antonopoulos, principal analyst at Nemertes Research and author of the New Data Center newsletter . "Instead, each business unit gets access to a slice of a server that allows it to meet internal [service-level agreements]. If you can transform the company culture to accept shared resources and lose some control and, in return, get better utilization and lower operational costs, then the question of outsourcing becomes a lot easier" to answer, he says.

Once IT reaches that point, it should have a good idea of what a specific application costs to deliver, in terms of reliability, availability and performance. Then it can structure SLAs appropriately and make an informed decision about whether outsourcing makes economic sense.

"If you can't measure it, you can't outsource it," says John Pierce, global solutions architect for outsourcer Patni Computer Systems. "The first question I ask when people are considering outsourcing is, 'Do you have a chargeback system?' If they say no, I think 'Oh boy, this is a disaster waiting to happen' because that means the users have no idea what this is costing them. How can an outsourcer be expected to control costs when IT doesn't even know what they are?"

In Pierce's experience, that means a company oftentimes finds itself needing to commit to a six- to nine-month engagement to determine costs, service levels and baselines before it gets to the outsourcing stage, he says. Only when those aspects are understood should enterprises lock into long-term outsourcing contracts, he adds.

A good starting point

The best starting place for organizations that already view IT as a service are easily measurable chunks of the environment, such as storage. "Storage is easier to transform into a service and therefore it's easier to outsource than other aspects of your infrastructure. There are complexities - an online megabyte is different from an offline megabyte - but they are nowhere near the complexity required for a server to deliver a specific application. Storage does not require application updates and operating system patches and so on."

Starwood's Conophy recommends starting with functions such as help desk, operations or network management, while Kaiser advises carefully considering the complexities of each chunk. "We've looked at outsourcing [human resources], for example,"Kaiser says. "In our case, we have to consider that we are a public sector organization that has a heavy union environment, with 26 bargaining units in the city. And that means there are certain HR functions and requirements that may not be as easily outsourced as in the private sector. But components of it - like payroll - might make sense."

Staffing issues

Staffing concerns often serve as an impetus, as was the case for Kaiser in his decision to outsource the city's infrastructure to Unisys. Rather than hiring high-end, expensive expertise, Kaiser can rely on Unisys to leverage its larger staff and resources. The result, he says, is better service.

Much of his operations staff went to Unisys in the deal - a decision Kaiser says was the best for the city and the strict technologists. "We're not in the technology business anymore," he says. "We need people who are focused more on business needs before they even think about technology. If you want to be in the technology business, go to a technology company. Go to Unisys."

The staff now is smaller, yet happier, Kaiser says. "In the past, we made a process faster through the application of technology, and we didn't look at the process and see if it made sense. But now, we focus on making those business processes more efficient. Plus, the staff can focus more on how the networks and infrastructure can best serve the business process, without having to worry about handling support tasks," he adds.

Breadth of expertise is definitely a reason to consider an outsourcer when moving to next-generation technologies, Nemertes' Antonopoulos says. "You need very broad skills, in terms of maintaining a highly heterogeneous environment and in understanding the technology's relationship to the context of IT as a service," he says. "So it's not just the technology, but how it affects the applications that are running on it, and what the dependencies are between that and the servers, and the network and everything else."

Most organizations don't have the resources to attract, maintain and manage people with such diverse skills, he says. "The outsourcing provider has the economies of scale to be able to ensure that it's not going to hire one person and then hardly use him because it has one Linux server."

Chicken or the egg?

Whether you choose to outsource or not, the key is to start migrating to next-generation technologies, experts say.

"The transition from a current state to the next-generation data center environment is one that we've seen overwhelming evidence pays for itself," Antonopoulos says. "So server and storage aggregation, consolidation and virtualization bring benefits immediately that will offset their costs. The upshot is that there's a very good ROI model for moving in that direction anyway - whether you decide to outsource or do it internally."

Cummings is a freelance writer in North Andover, Mass. She can be reached at

The IT complexity food chain

Binod Taterway, founding partner at outsourcing firm Blue Canopy, identifies four levels of complexity inherent in any IT organization.


Learn more about this topic

Outsourcing a corporate utility


Could you or your outsourcer handle a one-two disaster punch?


Outsourcing research center

Network World on Outsourcing newsletter

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2005 IDG Communications, Inc.