By Ashlee Vance, IDG News Service
November 15, 2002 | Keeping Celera Genomics should have been easy. It only took the largest technology merger in history, diverging product plans, and a fierce competitor for Hewlett-Packard Co. to lose one of its most important customers.
Four years ago, Compaq Computer Corp. -- now merged with HP -- secured its place among the top life science hardware vendors when Celera picked its AlphaServer systems to crack the human genome before any other group. The deal with Celera and the success of the project brought a flurry of press releases from Compaq in which company executives proclaimed the AlphaServer king.
Times change, though, and Celera has decided to give up its AlphaServers amid questions over HP's product roadmap and an aggressive sales pitch by IBM Corp.
Celera and Applied Biosystems -- both Applera Corp. companies -- announced in September that they have dropped Compaq's gear in favor of Unix servers from IBM and storage systems from EMC Corp. Both companies said they saw more of a future with IBM's pSeries servers than with the AlphaServer systems.
Uncertainty has been a problem for HP lately. The company took over the AlphaServer line after its acquisition of Compaq and plans to replace the servers with systems based on Intel Corp.'s Itanium processor in the coming years. Questions about HP's Alpha migration plans and other merger-related issues have dogged HP in recent months, creating an opportunity for other vendors, including IBM, Sun Microsystems Inc., and Dell Computer Corp., to close in on HP's turf in the life sciences market.
"I think there is a strong feeling of insecurity on the part of HP/Compaq users," says Ed Broderick, principal analyst at the Robert Frances Group based in Wappingers Falls, N.Y. "Call it uncertainty, call it unrest -- that's the situation."
HP’s Next Moves
HP's users face several dilemmas as they look at HP's product roadmaps. Over the next few years, HP will drop its AlphaServer systems and then abandon its PA-RISC processor-based servers as the company moves its high-end hardware to Itanium. Moving software to the still unproven Itanium chips, however, calls for a leap of faith and a lot of work from the end user.
In addition, HP will merge Compaq's Tru64 operating system with HP's HP-UX flavor of Unix -- a move that provides a long-term technology boost but more headaches for customers. With HP waiting until late 2003 to bring out large Itanium-based systems that could serve as AlphaServer or PA-RISC server replacements, customers in need of computing power now have tough decisions to make. Should they buy new AlphaServer systems to fill an immediate need, letting HP guide them to Itanium, or shift to another vendor?
"Why would you want to spend tens of millions of dollars upgrading to the same platform?" asks Mark Hall, research director of IDC's life science group in Framingham, Mass. "Alpha is not the future for HP. The future of HP is a year out or two years out from now."
For Celera, the future is with IBM and the 12 Power4 processor-based p690 servers it purchased.
"We were looking for a lot more than boxes and storage," says John Reynders, vice president of informatics at Celera. "We wanted to partner with someone in the IT space, so we could put our heads together. It was something a lot broader that we were seeking."
Celera is no longer solely focused on mapping the human genome. Instead, the company is tackling the therapeutics market and must use its servers for a wider variety of computing tasks. IBM's servers along with its AIX operating system allow users to slice up systems into partitions that can each handle different operations simultaneously.
In addition, IBM is bringing in its massive professional services arm to help Celera move data and software from Tru64 to AIX. Top researchers at IBM, coupled with the sheer size of its services group, make it a natural choice for this type of operation, Hall says.
Some users agree that IBM can fill all of their computing needs, giving them a one-stop shop, be it on the high-end pSeries systems or with clusters of lower-end servers like those used by the Institute for Systems Biology (ISB) in Seattle.
"We have a 64-node IBM eServer xSeries 1300 Cluster, with two microprocessors per node," says David Wilkins [cq], senior director of operations at ISB. "Our decision for IBM was based on an overall comparison of price and quality."
IBM beat Dell, EMC, and Sun for the ISB bid because the nonprofit group needed a cluster connected to a storage area network and sophisticated data-management software, Wilkins says. That said, Dell has come on strong with big life science wins in the past two months for large Linux clusters at the State University of New York’s University of Buffalo, the University of Pennsylvania, and Cornell University.
HP Holds Firm
HP contends that its services group is on equal footing with IBM and can help customers upgrade their AlphaServer systems or shift to PA-RISC-based or Itanium-based servers, according to Lionel Binns, worldwide life sciences manager at HP. "I don't think anyone, even IBM with its huge number of people, can claim to have the same experience we have had in complex and successful installations," he says.
Binns says that it did hurt to lose the Celera account and that the merger with Compaq has taken its toll, but does not see AlphaServer users abandoning HP en masse.
"There is no doubt that when two companies the size of HP and Compaq merge, it will have an effect on the environments of those companies," he says. "I am sure that we have lost some opportunities, and I am sure it has delayed some business for us. People get nervous, but we are coming out of that now."
Binns points to HP's recent win at the Sanger Institute for $22 million of new AlphaServer systems, which he says demonstrates how HP has tried to provide detailed plans of the planned shift from AlphaServer to Itanium systems to help those customers that want to stick with Tru64.
"We shared information with Sanger from very early on to help them build and architect a model to switch to the new processors at the appropriate time," Binns says.
"I've actually worked with Celera a little bit," says Ken Forman [cq], director of IT at Allos Therapeutics Inc., a Westminster, Colo.-based therapeutics company that runs on Dell servers. "I loved [the AlphaServers] and found them extremely reliable. They are a little higher-priced, but I felt that the reliability and speed was worth the cost. I was not surprised when [Celera] standardized on the platform, but I am surprised they moved off."
HP still has a long list of AlphaServer customers, including some of the most prestigious names in the industry. While HP offers these users a wide variety of technology choices, it needs to answer pressing questions about when to stick with current technology and when to make a transition.
"Some customers can wait to see what plays out with the new Itanium chipset," Hall says. "For Celera, it was clear they couldn't wait any longer." By contrast, the Sanger Institute was likely willing to save money now by keeping its software on the Tru64 operating system, as it plans to regroup with new technology down the road, Hall says.
IBM, however, has juiced its momentum in the life sciences community with the Celera win.
"HP is by no means out of the life sciences research game," Hall wrote in a research note. "But Celera was special. Though it no longer holds the world's attention as it did during the race to the genome, it still commands significant respect (and certain mystical acclamation). He who wins Celera wins a lot."
J. Craig Venter, the controversial pioneer who, as the former head of Celera, drove the first effort to decode the human genome, and now heads several foundations, told Bio-IT World he was "disappointed" to see the Alpha systems go.