Hi all,
In November 2011 I have deployed a single EqualLogic PS6100XV for a vSphere 5 setup and it all has been a very good ride sofar.
When I set up the volumes I made the choice to use thin volumes of 700 GB since we would stop filling up a volume on 80% actual usage (vSphere view). The reason behind this back than was that I didn't want to loose 8x 20% of 700 GB on "white space".
At this moment the thin volumes are closing in to their max (700 GB) and since I don't want any issues in de future I decided that I would turn things around to thick provisioning to we never could have issues with volumes going offline due to lack of space.
So this morning I started with my plan. I created an new thick volume "Vol05" and moved the VMs from the thin volume "Vol01" to the new one. When that was done I changed the configuration of "Vol01" by disabling thin provisioning on the volume in the management interface after which the volume showed as a thick volume, so far so good. I'm aware of the fact that this setting can be changed while using the volume but these kind of changes I rather perform on a empty volume for some reason unless this is really not necessary.
As I started to move back machines to "Vol01" the volume space still raised with every VM I moved on it which is currently on 645 / 700 GB in the management interface and 90 / 700 GB from the vSphere side and I started to doubt my own logic which caused my to open this post to ensure I'm not making any errors here.
The fact that the EqualLogic reports suck high usage is because it only can see which blocks have been written to in the past and isn't notified of VM removals (unless unmap in vSphere 5.1 is used) and thus the usage will be climbing to 100% after enough changes.
Q1: This is not in any way a problem when using thick provisioning right?
Q2: Monitoring the volume usage, in the above use case, directly on the EqualLogic is unnecessary as long as we are monitoring the vSphere datastore usage right?
With kind regards,
Henk Kraal