Hi,
DRS scenario: We have a 7 host’s in a fully automated DRS cluster, there are no resource pools in the cluster and almost no VM reservations.
vSphere ESX version: 4.1.0, 348481
DRS ‘Apply 1,2 & 3 recommendations’
The two largest host’s seems to get all the ‘attention’ from DRS, it seems that ballooning is preferred on the large host’s instead of loading VM’s to the smaller host’s that has free memory available.
Has anyone any thoughts of this scenario or some reading recommendations that we could dive into that explains it?
N.B. New hardware and more memory has been shipped
Are you using affinity rules? => could be that some of the "bigger VMs" are not allowed to be migrated to or from some specific hosts.
If you were to manually vMotion a VM from the busy hosts, would it work? => could be that DRS wants to move some VMs, but it can't because some vMotion requirement isn't met.
The resource distribution chart (cluster summary page > View Resource Distribution Chart) can also be helpful to get an overview of how are resources (CPU, memory) being used. Perhaps you can share a screenshot of that as well?
As for recommended reading:
- http://www.yellow-bricks.com/drs-deepdive/
- "VMware vSphere 4.1 HA and DRS Technical deepdive" / "VMware vSphere 5 Clustering Technical Deepdive"
No there are no affinity rules.
Yes vMotion will work from one of the busy hosts..
Great reading recommendations! Thanx..