2012-02-15 08:16 AM
I searched for an answer on my question but the things I have read don't clear this up for me.
I have the following situation:
aggr show_space -h aggr0
Now I already figured out that Snap reserve and WAFL reserve is already substracted from the Total Sapce (Avail).
However I wonder if apart from these reservations I should still keep a certain percentage of free space.
We are creating 1.3TB volumes with in it a 1.1TB lun. The luns are thin provisioned but they can have enough space to be fully filled.
The snapshots are hold on the same volume. I would really like not to size the existing volumes down as that might give issues with the snapshots(snapshots are deleted when not enough free space is available).
My question is, can 1300GB be allocated here? Or would that be a problem for performance, ...
When I look with df -Vh I see that the volumes on that aggregate are far from full.
Is netapp taking into account the actually used space or the full space that is "handed" over to a specific volume, even that space in the volume is not fully used?
So is the 1583 really usable, or do I still need alot of this space free to make sure I don't run into issues?
Note that dedup is active as well.
Thanks alot in advance!
2012-02-15 08:37 AM
The general rule is you leave 10% free space in your aggregate for performance reasons. So basically, of your 1.5 TB remaining I'd keep at least 150 GB free at all times.
One thing you may want to consider is turning off aggregate snapshotting. The only use they have is if you want to restore an entire aggregate (and thus all containing volumes) back to a point in time. These are independent from your volume snapshots. At least with vol level snapshots, you can mount previous versions as a snapshot and recover data. You can't do thsi with an aggr snapshot.
Turning off aggregate snapshots will bag you an additional 590 GB back to use for volumes and LUNs. Something to consider...