ADSM's philosophy on backups is that if the ADSM server has a good
copy of the data (a file) then the client does not need to send
another copy to the server. This leads some customers to have a
concern that one copy of the data could go bad, thus making recovery
impossible.
Here are the facilities in ADSM to minimize or remove
those possibilities:
1. ADSM V2 has a copy storage pool function. Once data enters the
server, you can copy it to many other "copy" storage pools.
Thus you can have backups of your backup.
2. Media reclamation. Once a piece of media is not fully utilized due to
old, expired data on it, the good data will be copied to a new piece of
media. In an operational system, you should see a regular turn-over of
media over time. ADSM is investigating time parameters to trigger these
reclamations as well.
3. Support for a wide range of tape and optical devices, thus giving the
customer a wide range of choices. Customers could choose cheaper
devices/media at the risk of media/drive failure, or higher cost
devices with substantially lower failure rates. Price, reliability,
capacity, and performance are all considerations here. This
ranges from QIC->4mm/8mm->DLT->3490->3590 for tape, and a variety
of optical units as well.
4. Ability to change the policies to cause full backups to different
storage locations. For instance, you may have mgmt-class-1 defined to
do incrementals, but on Saturdays change it to be absolute and to send
that data to a new storage pool. On Sunday you change it back to the
incremental model. User does not know the difference since this is
done centrally by an administrator. This of course caused more
data to be sent to the server, but may fit your needs if the above
options do not.
5. Media volume recovery. If a media volume goes bad, then it can be
recovered if you are using the storage pool backup feature mentioned
in item #1. If you are not using that, and this is backup data
(vs archive data), then when that volume is deleted from the
system ADSM will re-backup the necessary data to recover from this
media failure. Thus if file-a was on that volume and active, next
time a backup was done file-a would be backed up again. We call
this progressive incremental with a self correcting feature.
These are the primary facilities in the system to provide for data
redundancy and availability. We believe it delivers the right
combination of functionality and protection in a network friendly
manner.
Paul Bradshaw
|