At the National Center for Supercomputing Applications (NCSA) at the University of Illinois at Urbana-Champaign, not only is data growing astronomically, but it also appears to be increasingly useful.
In the past decade, the frequency with which researchers access data archived in the NCSA's mass storage system has increased from 18% accessed within the first two months to 50%; from 12% in the first six months to 28%; and from 3% in the first year to 18%.
In the early 90s, the NCSA's mass storage system was essentially a write-only archive, says NCSA technical program manager Michelle Butler. "At one point, transfer rates were so slow and cumbersome, it was easier for users to re-create the data than for them to retrieve it from the archive."
Managing data in the mass storage system is DiskXtender, HSM software from Legato. NCSA began using DiskXtender in 1992, and recently expanded its license from 500TB to 1.5PB to accommodate its year-over-year doubling of capacity. "The license gives us a little bit of breathing room," Butler says.