Research Center Grows Own ILM

ILM: an all-encompassing set of products and services from a single vendor.
That's the definition vendors who push ILM are pitching (see Tucci Touts ILM, IBMs ILM 30th Anniversary, and StorageTek Looks to Bag Buzzword).

Guess what: There’s no reason users can’t cobble together their own ILM tiers from different vendors. Tim Hunt, research computing support manager at the Fred Hutchinson Cancer Research Center in Seattle, did just that.

Hunt created his own ILM system from hardware he inherited when he came to Hutchinson two years ago, topping it all off with SATA drives and IBM Corp. (NYSE: IBM) Tivoli Storage Manager software.

Let's start at the top. The Hutchinson Center has more than 2,300 scientists doing cancer research, including three Nobel Prize laureates. Because of the nature of their work, they keep all of their data. Hunt says the center has 7 Tbytes of research data, and its capacity requirements expand 100 percent annually. He’s faced with backing up millions of files and a database with more than 21 million objects. And Hutchinson’s data growth rate will likely accelerate because of new initiatives.

“This isn’t just a fluke that we have 100 percent growth rate," Hunt says. "It’s been that way over six years.”Hutchinson stores primary research data on a Sun Microsystems Inc. (Nasdaq: SUNW) StorEdge 6320 midrange system with 12 Tbytes of capacity, linked via Fibre Channel to a Sun e4800 server. Two weeks ago, Hunt fired up a StoreEdge 3511 SATA system from Sun (see Sun Sings New Storage Song) as well. He also has tape libraries from Storage Technology Corp. (StorageTek) (NYSE: STK) and IBM (see Sun Guns Midtier Box). He uses Tivoli Storage Manager (TSM) to automate data migration across storage tiers.

Hunt says he also uses TSM to set thresholds for file systems. When a specified capacity is hit, the software migrates files that haven’t been used within a certain time period to the SATA drives and eventually to tape. The type of data helps determine how long files are kept on disk.

Hunt estimates he'll save hundreds of thousands of dollars by moving data off to cheaper SATA and tape instead of keeping it on enterprise drives. “Before, we were attempting to keep up with the growth by buying our way out of it,” he says, referring to the constant need to add Fibre Channel drives. “Automating the process lets us free up capacity.”

Hutchinson’s storage needs will only increase. Earlier this month, the center started an early detection initiative with extensive capacity requirements. The early detection group seeks to identify protein markers in the blood that can determine the risk, presence, and prognosis of cancer and other diseases.

“They have instruments that can look at your blood and say, ‘You might have prostate cancer in ten years,’ " Hunt says. “Each instrument holds 1 Terabyte of data, and there are four of them.”— Dave Raffo, Senior Editor, Byte and Switch