A recent report from London-based market research firm Silico Research finally puts some hard numbers on what industry observers have been aware of for some time now — while there is strong interest in distributed and grid computing from end users, adoption of the technology remains sluggish.
The report, “Analysis of computing strategies in the life science research and development process,” estimates that distributed and grid approaches made up only three percent of the $2.54 billion life science high-performance computing market for 2001, and projects this to reach seven percent of the $4.08 billion total life science HPC market for 2004.
The report — which distinguishes between “distributed computing” (splitting computational tasks among individual computers, generally located within a single enterprise) and “grid computing” (the broader application of this idea to multi-organizational projects involving the integration of computers, networks, databases, and applications) — surveyed 49 senior executives in life science research organizations to assess the current demand for new computational approaches. Of those surveyed, 69 percent said they currently use Linux clusters, while only 46 percent use distributed approaches and 23 percent have tried grid computing (see table on p. 6 for academic/ industry breakdown).
Top reasons cited by those who are not deploying these technologies were the stability of the technology and the unproven nature of the technology. Of grid computing, the report states bluntly that “the level of enthusiasm is overdone … and it will take at least three to five years before it delivers significant levels of processing power across organizations.”
Nevertheless, the increasing computational demands of biological research promise to secure a small but growing market for grid and distributed approaches over the next few years.
Putting the Technology to the Test
United Devices, an early mover in the push to sell the life sciences on distributed computing, has yet to secure a biotech or pharmaceutical customer, but that hasn’t tempered the company’s enthusiasm. Dave Wilson, vice president of marketing and business development at UD, said a deal with “the first major pharma to adopt edge-distributed computing” should be closed within a month.
Additionally, the company hopes to demonstrate the effectiveness of its technology in a new Internet-based research project it’s launching this week to identify key molecular compounds to fight the anthrax toxin.
The project — a collaboration between UD, Oxford University, the National Foundation for Cancer Research, Intel, and Microsoft — builds upon a similar project UD and Oxford have in place to screen small molecules against protein targets for cancer research.
Wilson said that while the cancer project piqued the interest of the life sciences sector, the anthrax project is closer in scope to the needs of the pharmaceutical industry. In addition, it offers an obvious and immediate purpose in light of the recent anthrax scare in the US.
United Devices estimated that it would take a pharmaceutical company several years to screen the anthrax protein against the 3.5 billion molecular compounds selected for this project on a typical supercomputer, while it expects its Web-based project to take around two weeks.
Acknowledging that most pharmaceutical firms wouldn’t want to access PCs connected via the Internet outside company firewalls to conduct a similar project, UD estimated that a corporate network of around 10,000 desktops would take about six months to complete the same task, while 40,000 desktops would take about two months.
Wilson said the company plans to turn over the results of the project to the US Department of Defense for further research and development.
Growing Interest from Big IT
Intel and Microsoft’s involvement in UD’s latest effort is only the most recent sign that the big IT vendors also hope to get a piece of the distributed computing market for life sciences research. Last month, Sun Microsystems said it was providing the infrastructure to support a similar effort by Sengent to identify small-molecule drugs to treat anthrax, smallpox, and other diseases caused by biowarfare agents [BioInform 12-17-01].
Sun is also hoping to market its Grid Engine software to the life sciences market. Cognigen, a Buffalo, NY-based company that provides data analysis and consulting services for pharma and biotech, recently began running Grid Engine on its Sun servers, a move that saves the company “an hour per day per scientist,” according to Duncan Ross, director of IT.
IBM, meanwhile, announced it was contributing hardware to the North Carolina Bioinformatics Grid in November [BioInform, 11-19-01].
Even Cray has thrown its hat into the grid computing ring. In December, the company donated a Cray SV1 system to the National University of Singapore’s Bioinformatics Center, which plans to plug the system into the Asia-Pacific region BioGrid. Tin Wee, an associate professor at the National University of Singapore, said the effort would be the first to port the Globus grid software to a Cray machine.
If anything, these stirrings from larger IT companies should be a signal to smaller firms that they’re on the right track in marketing their technology to the life sciences.
“As with any nascent technology, there are still a lot of people sitting on the sidelines,” said UD’s Wilson. “But once you get one or two or three picking up and moving forward it will add credibility to the technology and will also apply competitive pressure to others.”
While the sky may not be the limit for eventual adoption of distributed computing technology, Wilson said he remains confident that if it’s going to take off anywhere, it will be in the life sciences.