2014-06-01 02:27 PM
Our company just recently implemented 8040 ha controllers in our primary and dr locations and I'm hoping i can get some ideas on how folks have snapshot/vault schedules laid out. We are doing a migration from our old platform (Dell Equallogic) and our current requirement for data retention was 3 days. This is mainly due to the space constraints and also the way that eql handled replication. With dedupe and a better replication model thanks to NetApp, we can look at a longer retention period.
Currently I have configured volumes to do a local snapshot at 12am with the snapshot label of "daily". I have a vault relationship with the DR side and a policy schedule that will keep 7 copies of the "daily" label at a vault schedule of 1am. Here are a few questions i have:
Thanks in advance
2014-06-03 09:06 AM
Depends on the application, business requirement, RPO, legal requirements etc.
Yes, there can be a CPU impact so it's best to stagger snapshots and snapmirror/vault updates if possible. The general recommendation is to avoid taking snapshots on the hour, as DataONTAP does a lot of internal processing around that time. I try to schedule my daily snapshots to be 00:05 or 00:10. Also try to avoid de-duplication jobs, if any. It's best to schedule de-dupe jobs and then snapshot after they complete in my experience.
The SnapVault best practice guide is pretty old (2012), I'm not aware of a newer one but the principles are similar: http://www.netapp.com/us/media/tr-3487.pdf .Also check out the DataONTAP documentation for the version you are running. This is the Data Protection Guide for cDOT 8.2.1 https://library.netapp.com/ecm/ecm_download_file/ECMP1366833