April 1, 2008 | Geospiza's choice of Isilon as its primary storage partner might have something to do with the fact that both companies are based in Seattle. "It's very easy to go have meetings," jokes Todd Smith. More importantly, the Isilon system is very easy to scale. Data centers are handling thousands of Terabytes, with several staff responsible for ensuring the systems are working. "With the Isilon system, you need fewer storage administrators," says Smith.
The low cost of management is echoed by Matthew Trunnell, an IT architect at the Broad Institute: "We have a lot of storage in our environment, but we don't have a lot of FTEs to manage it all. Some storage requires more hand-holding than others." Trunnell says the management overhead for the Isilon cluster essentially amounts to just "one big file system."
Most of the Broad's network-attached storage consists of Network Appliance file systems. But as the size of the Broad's computational cluster has grown, "we've seen more and more frequent cases where jobs that are spread wide across our computational cluster can easily overwhelm the NetApp filer."
Trunnell's team evaluated several new storage architectures, including parallel file systems, but preferred the Isilon. "We have a diversity of applications, and it's very difficult to anticipate what the next demanding data access pattern will be," he says. Isilon allowed the Broad "to stay with something quite general purpose," without being locked into a vendor-specific file system that required client-side changes in the computational cluster.
Trunnell adds that, unlike other technologies, "The Isilon architecture scales linearly in terms of performance and capacity, since each little brick of storage includes storage processing and front-side bandwidth capabilities." --K.D.
Return to main article.
This article appeared in Bio-IT World Magazine.
Subscriptions are free for qualifying individuals. Apply Today.