Hybrid Unity 400 and VMWare Automatic Space Reclamation on VMFS6
Does Dell/EMC recommend using VMWare's VMFS6 Automatic Space Reclamation. If so, what setting to use, Priority, Fixed, etc. Does a Best Practice document exist?
The article was shared with me. The only issue that the article cites is possible array wide performance degradation. It looks like this is related to the rate at which UNMAP commands are issued, which can cause internal queues to saturate. It does say that the performance issue would stop when the UNMAP jobs stop, though.
I was looking into your question, and I found this best practices white paper. It has some detail on space reclamation, so you may find a satisfactory answer in there.
Thanks for the reply. Yes, I saw this document. It only pertains to an all-flash Unity and only covers reclaiming space within the guest, not on the datastore itself.
Well, that's unfortunate. I'll keep looking to see if I can locate something applicable. In the worst case scenario, our normal EMC support person should be back next week (I believe). If we can't find a resolution by then, we'll at least have some additional resources at that point.
Thanks. Please let me know what you find. We are having problems with Auto Space Reclamation on other, non-EMC devices and I want to ensure the Unity (and actually the VMAX3) do not encounter problems.
I'll see if I can find someone that can provide a more detailed answer. Without internal KB access, I'd just be risking providing misinformation and that's really the last thing I'd want to do.
This was from an internal KB that I don't have access to, so I'm unable to provide much detail, but it did indicate that VMware Auto Space Reclamation shouldn't be used.
Dell-DylanJ
6 Operator
•
2924 Posts
2185
0
Posted August 13th, 2020 14:00
The article was shared with me. The only issue that the article cites is possible array wide performance degradation. It looks like this is related to the rate at which UNMAP commands are issued, which can cause internal queues to saturate. It does say that the performance issue would stop when the UNMAP jobs stop, though.