Hi! I use DRS Affinity rules at present with DRS groups etc. My question is if it is possible to get more advanced with the DRS group management? Is there any way to group VM's together based on (tags for example) rather than manually? So ideally I could get to a stage where I can mark a large number of VM's in some way (tag/naming convention etc) and have them part of a DRS group.
The objective is to tie large numbers of VM's to certain hosts for license reasons, so I'm sure this has been done before. Obviously a separate cluster would do the job but not the most efficient use of resources. I'm looking for a type of soft-partitioning almost (in software!)
Any thoughts or suggestions would be greatly appreciated as always!
Thanks and regards
Steve
One way to do this would be to put the selected VMs on specific storage and serve that storage only to specific hosts within the cluster - that way you wouldn't need affinity rules at all, and the VMs would still be 'limited' to the hosts that 'see' the storage...
SteveR123 wrote:
Obviously a separate cluster would do the job but not the most efficient use of resources. I'm looking for a type of soft-partitioning almost (in software!)
Any thoughts or suggestions would be greatly appreciated as always!
Thanks and regards
Steve
Hi Steve,
When you have a server licensed for a certain software, and you can run only that software on those VMs hosted out of it, how will you efficiently use the resources?
Do you mean you want to put other VMs on it which are not bound by these license ties? Just trying to understand the situation here.
Hi, yes we could also use the hosts in that group for other machines therefore not reserving specific hosts to specific VM's (cluster method). However, the 'marked' VM's would only ever reside on a certain group of ESXi hosts. End result is only having to pay for (per-cpu) licenses on that group of hosts.
It's something we do at present using the standard DRS rules but I need to be more dynamic with the rule set.
Thanks
Steve
Let me guess - Oracle??
Actually No, not this time. We do use affinity rules for a small number of machines that run oracle which works well. This is apparently to cut support costs for RedHat, as we have a large number of them the usual manual DRS groupings doesn't cut it.
One way to do this would be to put the selected VMs on specific storage and serve that storage only to specific hosts within the cluster - that way you wouldn't need affinity rules at all, and the VMs would still be 'limited' to the hosts that 'see' the storage...
That's not a bad idea at all! Thanks for the suggestion.
Steve