2008-12-18 08:04 AM
Hi guys, can you give me a hint how to solve this:
We’ve got twelve esx-servers connected to our fas6030 metrocluster (184.108.40.206P1). On each node of the cluster there’s a 3tb volumes shared via NFS.
We have migrated the virtual machines from an old esx-infrastructure (on non-netapp san) to our current nas-infrastructure without optimizing the vm’s before. For that reason, the used snapshot space is more than 20% with a configured retention of one week (How much is it supposed to be on an optimized datastore?)
In the christmas holidays we’re going to change some thing on that infrastructure.
First of all, we’re changing the backup from ndmp “weekly to tape” (four weeks retention) to snapmirror “daily to disk” (also four weeks retention). The destination is a fas2050 (also 220.127.116.11P1), so the destination does not support the transfer of deduped data because it can not handle a 3TB volume due sis-limitations.
As a second thing we would like to optimize the virtual infrastructure and the hosted machines for the snapshot copies. Because we have migrated the vmdk’s directly from the old infrastructure, there are some machines writing logs, dumps, pagefiles,…. I have already asked for a solution to find the missconfigured machines () but I don’t know how I should use esxtop to get stats over one week to spot the snapshot-blowing vms. Can you give me another hint?
What would you recommend to change on the guests (mostly windows)? Thought about creating a separate datastore without snapshots for the pagefiles (nfs), a volume for database dumps (cifs) and a volume for logfiles (cifs). What are you doing to hold the overwrite rate low?
And now the final step: We would like to enable sis on the datastores to save the space-usage on the expensive fc-disks. I’ve read that NetApp recommends removing the snapshots before starting the deduplication. Does this just apply for the initial or for every scheduled process? Should I dedupe every night? Because the blocks are changing every time the process start, how does it affect the snapshots as we want to keep them for 4 weeks?
I know, a lot of questions, but may you can give me some tipps before I’m doing something wrong :-).
2008-12-29 05:52 PM
Each time you take a volume snapshot, every block in that volume is taken out of the active volume, and cannot be deduplicated (i.e. released to the free pool) until the snapshot expires, even if the block is a duplicate block. There are two ways to get around this:
1) Dedupe first, then take your snapshot.
2) Wait for snapshots to expire, at that time you'll get your space savings back
Important to note that you can't "hurt" anything with any combination of dedupe and snapshots, its just a matter of when you see your savings. Thare is much more detail in TR-3505, this doc can be found here in the community or in the library on netapp.com
Hope that helps