ESXi Cluster Performance Metrics and Host performance metrics discrepancy
I am trying to decipher the reason why the cluster peroformance metrics for CPU and Memory differ by about 10 - 15% of that of hosts values combined or average. We have one cluster with 4 hosts running esxi 4.1.
When I select CPU performance (counter: usage %) of cluster for the last hour or last few hours or last day, I see the utilization of hosts averaging 40 - 45%, but the cluster hardly crosses 32%. Similar observation is made for memory consumption as well. I am aware that if I am trying to compare days vs weeks, there might be datapoints taken at larger intervals and the results might not be that granular, but in an hourly view I think the cluster view should be close to the actual utilization on the hosts.
What am I missing here? I tried various counters but for cluster view, if I do custom there aren't many counters available. Any help would be appreciated.