Francois> I have a number of volumes/qtrees on my primary filer. I am Francois> busy setting up snapvault on a secondary filer (both have an Francois> equal number of disks available to them).
Is the secondary in the same data center, or is it across a WAN link of some type? That will change things alot...
Francois> I was wondering which might be better in practice: a Francois> one-to-one relationship of volumes and qtrees on the primary Francois> and secondary; or to create the requisite number of qtrees Francois> on the secondary inside a single large volume.
It depends. Running our SVs across a WAN link, we found that if you have a destination qtree which is busy in a Volume, all the OTHER qtrees which are being SnapVaulted to that same volume will have to wait until that long running SV completes.
So we've actually split out larger source qtrees off to their own destination volumes to get around this problem.
Francois> At the moment, I have 8Tbytes configured (4 volumes, 2 Francois> aggregates) on the primary (including snap reserve). I have Francois> about 5.5Tbytes of data to migrate onto the primary filer.
Hmm... I'd probably got with a 1:1 mapping, making two destination volumes and then filling them with the source qtrees. If you have any particularly large source qtrees, breaking them off into their own volume would be a win I suspect.
But again, it all depends on what type of network you have between the filers. If they're next to each other and on a GB LAN, then don't worry. If they're across a WAN link, you might want to think about doing your initial snapvaults to tape, then shipping them to the remote site and reading them in, then doing the catchup over the WAN.
John John Stoffel - Senior Staff Systems Administrator - System LSI Group Toshiba America Electronic Components, Inc. - http://www.toshiba.com/taec john.stoffel@taec.toshiba.com - 508-486-1087