[VMware] RAM Allocation

[CHiLL]

What's the best practice for RAM allocation when it comes to vSphere HA / failover if one host dies? At present I have ensured that each host is using HALF OR LESS of it's installed RAM. That way should one host fail, it can fail-over to the working host and have no performance impact. However that leaves a lot of unused RAM on each host. If I increase the RAM usage per host, then should a host fail, the working host would become over-committed.

 

I'm wondering what other people do in this type of scenario?

[Steve21]

I mean the short answer is will said downtime of servers impact you enough to worry while fixing the host?

 

We have hyperV set to only allow a 50% limit (e.g. 1 spare host "worth" of RAM at all times) so we can always migrate them over if needed.

 

But if you have some services, lets say SCCM that you can have offline for a week with no real issues you could just set that to power off etc

 

Steve