IGPP Digital Library, year 2002 – current

The IGPP Digital Library is a large mass storage system accessible 24/7 by computers within Scripps Institution of Oceanography (SIO). Any file within the 18TB system should be accessible within 2 minutes of being requested. Access to the Digital Library from outside SIO is through SIO work group computers and not through direct access to the library.

The current Library is well suited for all archival datasets regardless of file size or number of files, though extremes should be discussed with the system manager. While the physical storage is tape, all files appear to the user as disk files. Files are accessed from the user's Unix computer via NFS (Network File System).

The library is controlled by a software package, SamFS (Copyright LSC, Inc., now a Sun Microsystems company), that tracks the location of each file and automatically retrieves it from a Qualstar jukebox with 6 Sony AIT-3 tape drives. The present jukebox configuration has 180 AIT-3 tape slots, each tape holding 96.6GB. The jukebox can be expanded easily and cheaply to 360 tapes.

SamFS has been configured into 6 separate NFS filesystems within a single 35GB cache disk. The cache disk is used as an intermediary staging area so that access to the tape drives is in large chunks, which prevents frequent start/stops on the tapes.

SamFS has been configured to write every file to two different tapes. The intent is to remove one copy of each dataset from the system as it approaches 18TB since two copies of everything means the system only has 9TB capacity. The second copy is stored in a different building for safekeeping (no data were lost in the great IGPP flood). File redundancy within the jukebox is nice if one copy of the data is on bad media, the system automatically uses the second copy. This has happened only once in over a year.

Contact phenkart@ucsd.edu for further information.

Picture of the outside of the Qualstar

Picture of the inside of the Library.