2010-03-15 07:22 AM
Source: DS14 144GB
Destination: DS4243 450GB SAS
Data: VMware data with deduplication on with SMVI snapshots
When considering SnapMirror for migrating data from the DS14 to DS4243 (for that matter, between disparate disk size aggregates), I have couple of questions around best practices
a) What are the dedupe recommendations/best practices - turn it on on the destination while SM is running and set it to auto? or perform dedup after SM baseline is done?
b) Since the # of drives and size of the drives are going to be different on Source and Destination, is it a best practice/recommendation to run wafl reallocate on the destination? If so, I am guessing that the dedup should be off?
c) Given the block transfer nature of VSM and b), should we use QSM instead? any preference?
Looking for an optimal order of things.
Thanks for your help in advance!
2010-03-15 01:42 PM
a) Run de-dupe first on the source, establish VSM relationship, do the baseline & normally there is no need to worry about destination
b) Reallocate shouldn't be needed, but here is some interesting reading about swizzling & deswizzling (which should run automatically in the background):
c) To me VSM is an obvious winner
Have a look at TR-3505 though (http://www.netapp.com/us/library/technical-reports/tr-3505.html) - dedupe & SnapMirror is nicely described on page 24 and further.
2010-03-15 02:15 PM
Thanks for the responses. I am still a bit unconvinced about b), since I'll be moving snapmirror volumes between aggregates of different types and sizes. I've always been a fan of VSM but want to ensure that the data will be spread equally among all disks - if I need to run reallocate, I can. Will read up more and update this post.