=> On Fri, 24 May 2002 11:22:01 -0400, "Seay, Paul" <[EMAIL PROTECTED]> said:
> Yes, it will require all the inactive files to come back. But, the DR > vendor can provide enough disk. They cannot provide 30 tape drives. So > with about 8 tape drives restoring the pool in about 18 hours (estimate for > 5TB), we will then be able to do parallel restore of 180 servers. Here's a method that does not require all the inactive files to come back: Instead of a really-huge DISK pool, use that space for a really huge FILE devclass. Instead of RESTORE STGPOOL, on a per-storagepool basis, do a GENERATE BACKUPSET to your file devclass; this may be on a per-node granularity, or if you've got some really large filespaces, you may want to break them out. You can run as many of these as you have primary-tape devices, and you sidestep the possibility of your tape running faster than your network, and having to stop and re-seek; it's tape to "local" disk. You can start generating backupsets as soon as you are aware of the failure; actually _reading_ those backupsets will run efficiently up to your network capacity. You also dodge a bunch of TSM database contention. - Allen S. Rout
