Converting Serial to Parallel Database Loads Steven Lembark, Workhorse Computing
Track: System Administration and Infrastructure Date: Wednesday, February 05 Time: 4:45pm - 5:30pm Location: California Ballroom C
Merck's bioinformatics databases used to be loaded using shell code that ran each database sequentially. This frequently overran their load window and had serious restart problems -- mainly having to hack the code for restarts. Their system was rebuilt using Schedule::Depend, which now runs in a third of the time and can be fully automated. Lembark describes the environment and examines Schedule::Depend, specifically the alternatives it offers for data loading.
