URL : http:////docs.vmware.com/en/vRealize-Operations-Manager/7.0/com.vmware.vcom.core.doc/GUID-871D6B56-5...
Topic Name : Rightsizing
Publication Name : vRealize Operations Manager 7.0 Help
Product/Version : vRealize Operations Manager/7.0
Its not indicated what sample of time is used to analyze and determine that a VM is undersized or oversize. Does it look at the last week? Last month? Last year?... so it's based on how long.
Recommended size is generated from the capacity engine and is based on all of the data it has processed. The engine looks at the projected utilization in the future (based on the Time Remaining setting in the policy + 30 days). You can tweak the time period by adjusting the Time Remaining setting in the policy.
Oversized/Undersized VMs definition is based on the recommended zise. in vROps 7.0 the recommended size is produced by the capacity engine and is based on the projected utilization, particularly it is the max projected utilization during upcomming 120+30 days. The default 120 days is a configurable setting in the policy ( TR green zone).
Since vROps 6.7 the capacity engine is forward looking and is based on the forecst (unline the older versions, where it was backward looking).
I have a similar question. Where is it the policy defined? Where can the algorithm be changed? My concern is about whether or not the policy might be too aggressive when it comes to pulling away resources from VMs. I am interested in not only about the period of time it uses to determined VM sizes but what percentage of "iddle cpu or memory" (not sure if these are the actual metrics used) it uses to consider that a VM is wrongly sized. I can't seem to find any documentation on this.
Thanks in advance