Storage Efficiency

[ECEDERGREN]

we are in the process of moving data from FAS 2240 7-Mode 8.2 to FAS 8020 Cdot 8.3. This is all CIF shares.  On our old 2240 we are getting some realy good savings with Dedupe & compression. I'm using Robo copy to move over the data. Once the data is copied overto the  new FAS 8020 I'm getting no where near the savings that I was getting on 2240. It's the same data on each of the volumes I could live with and understand a small veriation but this is huge differance. It's also not consitant across all of the volumes that I have copied over some are getting better savings.

FAS2240

FAS8020

[bobshouseofcards]

One obvious question - were the space efficiency options enabled on the target cDot volumes before you started the copies or after?

Bob Greenwald

Lead Storage Engineer, Consilio LLC

NCIE SAN Clustered, NCIE Data Protection

[ECEDERGREN]

It was turnned on prior to copy

[bobshouseofcards]

The other obvious question then - what is the snapshot schedule like?  Do you have the volumes running regular snapshots, and how long are they being maintained?

Remember that space determined to be saved due to a compression/dedupe run is logically "deleted" from the volume.  However, if you were snapshotting the volume before/during that roughly 5 day initial space efficicency run, all that deleted space is still in the snapshots so you haven't "saved" it yet.

As the snapshots expire, or as you delete them yourself, additional savings will be had.  For instance, on a particualr volume I have that is already at 49% dedupe (12TB volume actual, 11TB saved to date, 23TB logical) the most recent run will save an additional 0.7TB in space efficiencies, but that space won't show up for two weeks because it's working through the snapshot retention times.

What does the "vol snapshot show" list for the volumes of interest?  Are there snapshots that hold a lot of space?

Hope this helps.

Bob Greenwald

Lead Storage Engineer, Consilio LLC

NCIE SAN Clustered, NCIE Data Protection

[ECEDERGREN]

Bob,

Thanks this was great information. Curently I'm not running any snapshots copies but I am snapmirroring to a DR site. Currently only have this running once a week.

I'm thinking that it's best to just delet that snapmirror relationship, after I get all data copied and it's been deduped & compressed, then set it up again.

Or just setup the snapmirror to run more often

Would love to get your thoughts

[bobshouseofcards · Accepted solution]

The Snapmirror relationship holds a Snapshot based on when it last ran, so yes that could easily be hold the space from the dedupe process. Guessing you run the mirror update on the weekend? I'd just wait until Monday and see where you are. You can confirm how much data is pending for delete by checking the size of the Snapmirror snapshot to see how different it is from the current volume.

Unless the Snapmirror is still doing the initial update and it still has a decent amount to go? If so given the amount of data you may possibly save I could see killing and re-establishing the mirror now that the first dedupe cycle has run.


Hope this helps you.

Bob Greenwald
Lead Storage Engineer, Consilio LLC
NCIE SAN Clustered, NCIE Data Protection

[ECEDERGREN]

Thanks for your help, after the weekend I did see some more savings but stil wasn't what I had on my other system. So I pretty much started from scatch. Deleted Snapmirror relationship, deleted volumes. Recreated volumes, recopied the data and let the schedualed dedupe and compression cycles run. So far on the volumes I have done this I'm getting at least equal amount of savings from Dedupe and compression.

[DocMikeWoods]

Bob,

Thanks for your previous post on deduping data and saving storage. I was wondering other than compressing snapshots if you have any additional tips on freeing up space?

Thanks,

MW