HPE StoreVirtual Storage / LeftHand
cancel
Showing results for 
Search instead for 
Did you mean: 

Question on Consumed Space size oddity?

SOLVED
Go to solution
Massoger
Occasional Visitor

Question on Consumed Space size oddity?

I've got 2 P4500G2 in a cluster with 24 600GB Drives.  Total space is 8.7TB, running Raid 10.  I've got 9 volumes all THIN provisioned.


1 Volume in particular, assigned to an MS SQL server is reported as 4TB, with the consumed space showing 4.81TB.  The odd thing is that the total data usage on the VM itself is only reporting a total of right at about 1TB.  So where are the other 3.81TB coming from that is eating up all the space on my SAN?

 

I've read every document I can find from HP on Thin/Thick provisioning and all sorts of forum threads and nothing that is supposed to be happening seems to be happening.


Running version 9.0.0.1 on all pieces execpt for CMC which is 9.5.

 

The SAN is now over 98% full because of this one Volume...any help is greatly appreciated!!

Thanks,
Dave

2 REPLIES
Jitun
HPE Pro
Solution

Re: Question on Consumed Space size oddity?

Volume Size 4 Tb

Network RAID 10

Total Max possible Consumed Space (without Snapshots) - 8 Tb (Thin/Full)

 

So, as you have 4.8Tb consumed Space, you have used up 4.8/2 (network RAID 10) = 2.4 Tb space in  the volume.

 

As you say, your Host is able to see 1Tb of consumed space on the volume.

You should be asking where has  2.4Tb - 1Tb = 1.4 Tb space gone ?

 

Please remember that once a block has been assigned to a volume in SAN(Cluster), the block cannot be reclaimed by the SAN(Cluster) till the Volume is present in the Cluster.

 

So if you had, at any point in time, used up 2.4 Tb space from the Cluster for the volume, the the Cluster would have allocated 2.4Tb space for the volume, now as it is a Network RAID 10 volume the volume would use 4.8 Tb of space.

 

Once it is consumed, the blocks are not released back.

That is, once data is written to the volume and then deleted, SAN would not know if there is data in the block or an empty are.

 

 

The only way to reclaim the un-used blocks is to delete the volume and recreate it.

 

98% filled up SAN is not Good.

Recommendation - Delete any un-necessary snapshots.

 

Please move data out of the volume, create a a new volume and copy the data back to the new volume.

Ideally if I had more space on the SAN, I would have create a new volume(thin Provisioned) , assigned to the Host and copied the data to it.

 

Hope this a fair enough explanation.

--------------------------------------------------------------
How to assign points? Click the KUDOS! star!
Massoger
Occasional Visitor

Re: Question on Consumed Space size oddity?

Perfect explanation!  I was under the impression that after data is deleted, the SAN would release that space back to be reused elsewhere.  It seems very difficult to have to delete and recreate volumes to free up space that is technically not being used.

 

It was our DB server that had been written to with log files to consume the 1.4 TB of space, that is now not being used any longer, but in fact eating up all the space on our SAN, so we are in fact having to recreate volumes and move the Databases and go through a whold bunch of hassle in order to free up SAN space.

 

Do you recommend Thin or Full provisioning for Volumes containing SQL Databases?

 

Thanks again for taking the time to post the explanation, it makes total sense now.

Dave