Starwood Hotels and Resorts, the White Plains, N.Y., hotelier, is in the midst of migrating from a mainframe infrastructure to a next-generation computing environment built on Unix and Linux servers. CTO Tom Conophy expects that its next-generation infrastructure will save Starwood more than $20 million a year. Conophy discussed his outsourcing strategy in a recent interview.
Starwood Hotels and Resorts, the White Plains, N.Y., hotelier, is in the midst of migrating from a mainframe infrastructure to a next-generation computing environment built on Unix and Linux servers. By August, the company will be using its new infrastructure to support more than 150 Java 2 Platform Enterprise Edition -based Web services for use by its 733 Sheraton, Westin, W and other brand hotel properties worldwide. CTO Tom Conophy expects that its next-generation infrastructure will save Starwood more than $20 million a year - in part through the savvy, judicious use of outsourcing - while making the company more competitive and better at serving customers. Conophy discussed his outsourcing strategy in a recent interview.
For the move to this next-generation environment, you kept control of application development while outsourcing almost everything else. How did you make that decision?
We know the costs of maintaining, engineering and supporting the network hardware. We outsourced [the infrastructure portion] because it's closer to a commodity and it's easier to make sure we have competitive pricing and so on. What we've kept in-house are the reservation and loyalty system applications and the development of those because they are more critical to the business. This is where the unknown areas are in terms of complexity, features and new functions. If you have a mature application base and you're just into maintenance, or if you're in an industry that's not going through change, then outsourcing may well be the most cost-effective way to do it. But where we see the biggest savings from a pure [total cost of ownership] perspective is in the cost of the hardware and the maintenance on that hardware, the software licensing and the managed services side. We built the business case around those three items, and saw that we could save potentially up to $20 million a year in this new platform, relative to what we started with.
So the outsourcing decision is about controlling what's core to your business while offloading what you can get more cost-effectively elsewhere?
Correct. Could you operate your own data center efficiently and in a cost structure that might be better than these large managed services companies? Odds are, yes. But with skills and resource constraints, can you achieve that in a midsize company like we are? If I was Bank of America, maybe I'd run my own data center because I'd get synergy by the size of my operation. But for Starwood, it comes down to whether we want to take on that burden. A well-run organization might be able to get the cost-efficiencies, but it's one of those things where the cost differential just isn't that much, so you might as well outsource.
After whittling down your outsourcer choices to IBM and HP, you picked HP. Why?
We selected HP primarily because of its overall flexibility. We knew that our business was going to change over time, and we wanted to make sure that we were not locked into a restrictive managed services contract where we'd be paying penalties because of system growth.
Also, we wanted to have IBM or HP augment our staff to co-develop and build out this reservation system, and IBM was less flexible about the ownership of intellectual property. Starwood was not going to give up the intellectual property, and HP did not have issues with that.
Did support of new data center technologies, such as virtualization and utility computing, factor into your decision?
It's a factor, but more as a future. We own our hardware, but if we were in some kind of a leased model where we rent CPUs for our high season, it might make sense. But right now, with the cost of hardware being what it is, and the computing power of the current chipsets, it's just not a challenge. If I need to add more capacity for my shopping engine, I throw a few more Linux processors at it.
And your IT staff is OK with the infrastructure outsourcing strategy?
[Yes, because] the internal staff is part of the core team, which we've augmented with HP resources. Now when we finish this big development push, the HP staff can go. But my core staff will remain to face any new challenges, and they won't be burdened by the more mundane [hardware] maintenance-type work.
Cummings is a freelance writer in North Andover, Mass. She can be reached at jocummings@comcast.net.