Hello:
I have noticed two basic approaches to parallel backups and restores
for Netware clients described on this forum.
One is from Dwight that suggests doing several concurrent backups from
one node using a different node name for each volume on the node,
thereby enabling you to do concurrent restores with no tape contention
(collocation must be on).
The other is from Andy Raibeck suggesting the use of separate storage
pools for each volume, but running in one process using one node name
(using include statements to spec a management class for each volume,
with each management class pointing to its own storage pool).
Can these two ideas be combined? Is there any reason I can't run
multiple concurrent backups on one node, with each session having its
own node name, and each session backing up one volume, each volume
having its own storage pool (management class)? If I understand this
correctly, this would enable me to run not only parallel restores, but
parallel backups as well. We have trouble getting our large nodes to
complete in a night, so I have incentive to speed things up in both
directions.
Larry Schiebel
University of Wisconsin
|