On Mon, 7 Oct 2002, rick pim wrote:

> Subject: [Networker] large filesystem restores
>
> environment: solaris 5.8, networker 6.1.1
>
> in order to speed up backups on a large filesystem, directories are
> arranged as follows:
>   /export/home/home0
>   /export/home/home1
>   ...
>   /export/home/home9
>
> this permits full backups to be spread over several nights. fine.
>
> last week we had to restore the filesystem. the person doing it used
> recover and simply started a single recover at /export/home. as could
> be predicted, this took a long time. the main bottleneck appears to
> have been the single restore thread using a single tape drive at a
> time. presumably we could have saved time by starting (say) two
> recovers at once, manually spreading them over the directories to be
> restored.
>
> the question is: is there any way to tell legato to (automatically)
> use more devices for a restore? is there any way to get recover to be
> a bit smarter wrt resource utilization?
>

Rick,

Most of the effort has gone into making dumps fast rahter than restores, since
most backups spend 99% of the time doing dumps rather than restores.

There is no way to tell Legato to use more devices except by starting multiple
restores.  You can still run into tape contention problems if both processes
are attempting to read from the same tape.  Depending on the number of files
in the filesystem, you can also run into a lack of virtual memory during a
restore.  If you are restoring from a single full dump, saveset recover is the
fastest way to restore.

Shelley

--
Note: To sign off this list, send a "signoff" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=