Hi Guys,
I am hopping to get a little clarify about the algo that evaluates the size of the VMs vs the CPU/RAM usage.
We have some very busy VMs that can be used 60-70%CPU during business hours but are completely idel after hours. Does vROPs count the idle time towards its calculation of sizing the VMs? Is it possible to time box its collection periods to only during business hours?
We are running VC6.7U3, vROPS 8.0
Thank you.
I do realize that there is a Workload Pattern thing, but it does not appear to be accurate.
vs the CPU usage graph for the same period
this is not 1 of the VMs that goes full tilt during business hours.
Its dependant on what is set in your policies, ultiamtely you should use it as a guideline and make your own decision on whether it meets the servers actual requirements.
Take a look at this: Rightsizing
"Oversized VMs" displays the number of VMs determined to be oversized based on policies previously set. A chart details suggested reductions in the overall number of CPUs and GBs of memory and shows the percentage of total resources the reductions represent
Just trying to read through the policies KB. Any hints on a particular policies setting i should change, under what heading its located?
It is using the standard data collection period (every 5 minutes).
check the workload automation policy element
Below blog provides everything you need to know about rightsizing
Rightsizing VMs with vRealize Operations - VMware Cloud Management
RickVerstegen
Had a look at the blog, i seem to be blind but i cant see where it says anything around sample limiting or time-boxing.
Are you able to give to say where it says how to do this?
T180985
How does this change the sampling or what time-box window it uses for statistics
Gr33nEye
As far as i know sample limiting / time-boxing is not adjustable. It uses the standard data collection period which is 5 minutes by default.
Is there a way to force the collection to look only at the active memory not consumed.
Based on the recommendations, and the VMs overview it is only looking at the consumed memory. This is always close to 100% of available VM so all the VMs need a RAM increase.
I've changed the base default policy to disable local for consumed and enabled local for Memory|Non Zero.
Will this force vRops to look only at the active memory usage?