Scientific Data on Demand – NERSC’s High Performance Storage System
The High Performance Storage System (HPSS) at NERSC can efficiently manage up to 37 petabytes of science data by moving information from high-cost, high-speed disks to low-cost, energy efficient tapes. Thirty-seven petabytes of information is equal to all the music, videos or photos that could be stored on 321,900 iPod classics filled to capacity.
When a user requests HPSS data, a robot arm (bottom left) finds the cartridge using a bar code to verify the data inside. The robot then mounts the carridge in a tape drive, and streams the data back to the user. Typical tape mounts take between 30 and 90 seconds to complete.
NERSC currently runs two different HPSS systems: an archive that stores scientific data files generated by researchers around the world who want long-term retention of their information; and another backup system that contains copies of information from various other sources at NERSC, so that user data can be restored in the event of a disaster. This image shows one row in one of the new HPSS tape libraries. The new tape libraries increase HPSS theoretical capacity at NERSC to 37 petabytes.
NERSC’s HPSS system currently holds about 59,000 tapes, each with storage capacities ranging from 75 to 1,000 gigabytes. The oldest science file in HPSS is from January 2, 1976. As of 2009, the HPSS archive system contained over 69 million files and more than 4 petabytes of data; the HPSS backup system contained over 12 million files and about 3 petabytes of data. The systems are Sun Microsystems’ StorageTek SL8500 Modular Library Systems.
credit: Lawrence Berkeley Nat'l Lab - Roy Kaltschmidt, photographer