After upgrading our vCOPs to the most recent version we are now seeing a large number of alerts on CPU or Memory stress and I was curious what exactly this is and how to resolve it? I review the VM's undersize / oversize and projection charts and from what I can see there may have been a single day that the stress occurred (or very few hours in that day) but the remainder of the time the VM's are showing over sized. This behavior is very common in our environment were a VM sits fairly idle or does very little work till and end of cycle occurs and then the VM works a lot. So far with the behavior our environment has I am having trouble really keeping our management happy with the product when they see that the product offers reports that both show a VM is stressed yet over provisioned and was curious if there is any recommendation on more of a fine tuning type procedures for the product?
Attached is an example of the undersize/oversize type charts we commonly see.