How to seed first SnapMirror w/o using network?

[kmbartels]

I will be deploying a FAS2020 at a remote site that has 2-TB of data.  I plan to setup and migrate from direct-attach in a Windows environment.  After the migration, I'd like to use SnapMirror to mirror my data back to central site on a FAS3140A.  The problem is bandwidth for initial sync.  The 2-TB will take about 90-days to move across a 3-mbit connection.  We're okay with the write updates, as these are only about 1-GB per day.  Any ideas on how to get the initial data over to the central site without physically moving the FAS2020?  Can we backup to a tape via NDMP and load the tape at the central site?

[adamfox]

An NDMP dump/restore won't do it.  However, assuming you are doing Volume SnapMirror, SnapMirror to Tape will work.  See the 'snapmirror store' and 'snapmirror retrieve' commands.  If you are doing Qtree-based SnapMirror, you'll need to look at lrep which should be in the NOW Toolchest.

Hope this helps.

[kmbartels]

Thanks Adam!  Yes, I plan to use VSM.  Do I need to be concerned about quiescing data before running 'snapmirror store' to tape?  I'm assuming that I can keep my volumes online after the store, as it will take a day or two to ship and load the tape at my central site.  I'm also assuming that the 'snapmirror resume' will pick up exactly where the 'snapmirror store' left off.  Am I correct in these two assumptions?

[adamfox]

You do not have to quiecse anything.  The snapmirror store is basically the equivalent of your snapmirror intialize.  The Volume can stay online while the store is happening and can say online while the data is being shipped.  At the remote site, the destination volume may have to be restricted during the snapmirror retrieve, but that's just like it would on initialize and doesn't affect anything on the source side.  Only after you've done the baseline should you set up the update schedule.

Hope this helps.