Some VMs are oversized, causing ballooning within resource pool. There are lots of memory counters in performance chart, what's the best way to estimate the "right" size of VM based on historical data?
Thanks.
any suggestion?
I think you need to use a combination of vCenter memory counters with in guest memory counters. You want the VM to have enough memory to handle spikes without the guest swapping. So take a look at active memory from vCenter and maybe page file usage and available MB inside the guest OS. I'm also curious to see what others say.
I'm thinking to use Granted (max) + Swapped (max) or VM configured size - vmmemctl (avg) - zipped (is there such counter?)
Memory counters at http://www.vmware.com/support/developer/vc-sdk/visdk41pubs/ApiReference/memory_counters.html