vSpeher 4.1 - Strange DRS behavior

[Kastigar12]

Hi all,

I have couple DRS and HA Clusters which are shown different behavior. DRS is set to full automate.

Main Problem is that in Cluster 3 one host is always above normal on memory utilization (>80%) but DRS does not apply any recommendations.

In Cluster 1 and 2 there is no host with cpu or memory utilisation above normal, but DRS migrate each 10-15 min some vm from one host to another.

I have tried to find some documentation which could explained such behavior but without effort.

Will be appriciate for any help.

Thanks.

[Troy_Clavell]

DRS may in fact think the cluster is balanced. If you check the summary tab of your cluster and the VMware DRS box.  You should see "Current Host load deviation"  is it green?   You may try and manually load balance your cluster, or set the migration threshold a bit more agressive if you are concerened.

[Virtualinfra]

Take a look at the below blog for DRS understanding, It depend on the migration threshold set. if its set to 3 and resource calcuation is done based on the available resource onthe host, if the cluster calculation say the it is in balance there wont be any migration happening.

http://www.yellow-bricks.com/2009/10/21/drs-deepdive/

Award point for helpfull and correct answer by clicking the below tab

[Kastigar12]

Thanks for all, really helpful.

So, I will see next time as one of hosts in a cluster is overloaded for this current host load standard deviation.

But one point I do not get some how. Could be that if one host from 3 is overloaded, but DRS mean the cluster is still balanced? Does DRS not see separate hosts but just a cluster?

It is so?

Thanks.