HSM (Hierarchical Storage Manager) System
General background
The HSM (Hierarchical Storage Management) System is an integrated
storage system consisting of disk storage (108 GB), optical disk storage
(208 GB), and tape storage (420 GB). When you save a file on the
HSM (/hsm/users), the file is almost immediately copied in a compressed
format to the secondary storage devices (optical disk or magnetic tape).
Files on the disk that have not been used recently may be temporarily removed
from the disk, by the HSM software, to make space for files in current
use. However, the disk's directory will make it appear to you that
the files are still on the disk. When you try to access a file that's
been temporarily removed from the disk, the file is automatically retrieved
from the secondary storage without any special action by you. Retrieval
times vary depending on the other system activity at the moment.
For example, the tape drive or optical disk drive may currently be in use
or the robotics device that loads and unloads the secondary devices may
be busy retrieving a tape.
The HSM system provides you with storage capabilities
much larger than are available on the conventional filesystems (e.g., /home/base/usrb,
/home/base/usrc). However, the typical read/write times for the HSM
system are longer than for the other filesystems. Technically, this
is because the other filesystems use UFS, a much higher speed data link
than NFS, which is used by the HSM.
Disaster recovery procedures for the optical and tape
backups assure a high degree of reliability in case of a complete system
disaster. (In case of a complete disk disaster, the entire disk can be
recovered from secondary storage devices. Furthermore, the compressed version
of the disk file is copied to two secondary devices for additional security.)
However, if you personally delete a single file or modify it, you will
not be able to recover the previous version yourself or through a request
to IT staff.
Appropriate use of the HSM system
Because of the technical issues described above, you would
only want to use the HSM system for the following situations.
-
Storing large datasets that have copied from tape or
CD, primarily to be used as input data.
-
Storing results from long computing jobs that generate much
output needed for archival purposes.
If you are producing a file that want to archive on the HSM
system, but the file is difficult to recreate, you may want to temporarily
store the file overnight on a conventional filesystem (e.g., /home/base/usrb).
Files on those filesystems are backed up by the Legato Networker software
and can be retrieved by the UNIX recover command if the file is
not too old.
Because of the nature of the HSM system to copy disk files
almost immediately to the secondary storage devices, the HSM system is
not an appropriate storage medium for constantly changing files.