Making a Huge BetBy David F. Carr | Posted 2007-07-18 Email Print
WEBINAR: On-demand webcast
Next-Generation Applications Require the Power and Performance of Next-Generation Workstations REGISTER >
University of Pittsburgh Medical Center had runaway growth in its server and storage infrastructure. Here's what it did.
Making a Huge Bet
It was big news in 2005. The University of Pittsburgh Medical Center that year made a huge bet on an IBM partnership that the two organizations hoped would not only alter how UPMC used information technology, but change the way vendors and customers work together and even transform the health-care industry.
With the $402 million, eight-year deal, UPMC agreed to lock in IBM, making Big Blue its primary server and storage technology provider. In return for that big commitment, IBM said it would provide generous discounts and, further, agreed to help with the full range of people, process and technology issues surrounding the overhaul of UPMC's computing infrastructure.
The primary aim was to whittle down UPMC's ongoing cost of operating its server and storage infrastructure. At the time, the hospital's demand for information services, as measured by requirements for additional servers and storage, was growing at about 20% per year. "We were being crushed by our own infrastructure," says Paul Sikora, the executive in charge of what UPMC calls its Transformation Project. "We saw increasing demand all around us, while at the same time systems had to be more reliable and run faster. We didn't see any light at the end of the tunnel for additional funding or staffing, so it became a question of how do we do more with what we have."
As part of the deal, the two parties set an ambitious goal of reducing UPMC's I.T. infrastructure spending by up to 20%.
UPMC earmarked the savings for its strategic initiatives, such as continued refinements to its electronic health record system. IBM and UPMC also agreed to work together on developing applications that they could jointly market to other hospitals and health-care firms.
In a prepared statement touting the deal, UPMC's hard-charging CEO, Jeffrey Romoff, stressed the importance of information technology to quality, to patient care and to bring "the latest in medical research to the bedside," and concluded, "IBM is the right partner for us, and the right organization to help transform the industry."
Still, before the parties shook hands, UPMC got advice from a couple of I.T. advisory services it consulted, Gartner and Forrester, that the deal might be a mistake—they thought UPMC might be giving up too much by locking itself into one vendor, rather than making it bid on each new purchase of servers, desktops and other equipment.
UPMC chose to go ahead anyway. "We decided we could take advantage of the aggressiveness we got from the vendor in a competitive bid, and make it for an extended period of time," says Sikora, who notes that there's enough flexibility in the contract to let UPMC buy from other vendors if the hospital believes their products are a better match. A case in point: UPMC went with Hewlett-Packard's ServiceCenter as part of the service management initiative within the Transformation Project. "We don't think we've lost the advantage," he says. "We think we've magnified it."
So, two years into the partnership, how's it going?
While it's not clear if the UPMC-IBM tandem will transform health-care delivery, the project is dramatically reshaping UPMC's processes and technology:
• UPMC is moving toward what IBM calls an On Demand computing environment, which is essentially another name for utility computing—the concept that computing resources should be pooled so that the business can draw more or less heavily on them as business needs change, with proportionately graduated pricing.
• UPMC is taming its once-rapid proliferation of servers with virtualization, meaning that many applications that formerly occupied physically separate servers now run in logical partitions, or "virtual machines," on a smaller number of servers. This cuts capital costs and administrative overhead. Because virtual machines can be created or reconfigured more quickly than physical computer servers, virtualization also fits into the On Demand strategy.
• The medical center is deploying a smaller number of operating systems and generally trying to whittle down the variety of software in its infrastructure for simplified administration.
• UPMC is standardizing service and support functions around the best practices of the Information Technology Infrastructure Library (ITIL).
And UPMC is tracking toward the goal it set in 2005 of a 15% to 20% reduction in the operating budget for systems infrastructure. So far, infrastructure spending has been cut about 13%. Of the total $122.6 million operating budget, UPMC expects to spend $14.1 million on infrastructure this year. That's 11.5% of the total, down from 14.7% in 2005.
Achieving those results required a serious commitment, according to Daniel Drawbaugh, UPMC's chief information officer. "Probably the easiest thing for a CIO to do is pursue this on a gradual basis, a case-by-case basis," he says. "But to really transform your technology environment, you need to approach this with a much more focused and enterprise-scope project." And while the server, storage and network infrastructure may not be glamorous, they are the fundamental pieces everything else is built on, Drawbaugh says: "The organizations want to have the applications, but you've got to have the infrastructure to support it."
UPMC is determined to have the best because, despite being a non-profit university medical center, it's also a very aggressive, entrepreneurial organization. Today, UPMC is widely recognized as a leader in health-care I.T. In seven of the last eight years, U.S. News & World Report magazine, which annually ranks the nation's best hospitals, put UPMC on its America's Best Hospitals Honor Roll. And the Bush administration has cited the medical center's eRecord system as a model for the kind of electronic health record other hospitals and health systems ought to be implementing to improve patient safety through better sharing of medical information.
The eRecord system is actually a suite of more than 30 applications, supported by more than 130 developers, and has been under development for years. In fact, UPMC traces its roots to an electronic medical archive system developed 20 years ago. ERecord currently reaches 13 hospitals, more than 150 outpatient facilities, 2,500 concurrent users and 3,000 desktops. Every tier of the application is configured for high availability, and the "safety critical" rating of this system means that it would have to be brought back online in no more than four hours even in the wake of a crippling data-center disaster. Although eRecord development is proceeding on a separate track, the Transformation Project helped UPMC figure out how to run it more efficiently, cutting the application's processor requirements approximately in half.