I have a had a couple of questions on when capacity calculations occur and if they can be changed.
vROps 6 is allot more flexible when it comes to capacity calculations. In vCOps 5.x when you made a policy change you had to wait 24 hours to see the results of that change.
For those who might be wondering when you make a policy change on a cluster it will automatically kick off a capacity calculation. It will then run it once every 24 hours for example make a policy change at 7pm it will then do a calculation at 7pm the following day.
This can be changed though modifying a file called capacity.property located at /usr/lib/vmware-vcops/user/conf/analytics/
In this file there is a value capacitycomputionperiod which by default is set to 1440 minutes (24 hours) change this value to say 60 minutes and we get a capacity calculation every hour, from the last policy change.
previously data point every 24 hours
After the change data point every hour
On a side note make sure to make this change on all the nodes in the vROps 6 cluster.
Cheers
Not sure how to email you a question. Scott, in your book, Mastering vRealize Operations Manager, you state in the Alerts section, that the wait cycle and cancel cycle are the “number of cycles we need to wait for the alert to be triggered”, but I can’t find out how long a cycle is? 5 minutes? 30 minutes? What’s a cycle? Thanks.
Hi Ron,
Cheers heaps for your comment for emails you can reach me at snorris@virtualiseme.net.au.
A cycle is by default 5 minutes it is the metric “collection” interval.
Thanks! It’s funny how the documentation continually references “cycles”, but I never could find out how long that was… I’m sure it’s mentioned at least once, but I missed it
Another question about capacity calculations. Is there any way to get the Density optimal VM ration calculation to use the “Average VM” size instead of the “Small” VM size? What I mean is, if you look at the Density report, it might say 80:1 is optimal, but if you do that math, that’s only if every VM built from now on equals the “small” VM size found on the Capacity Remaining page… I’d love to change that calculation to use the number of “Average” VMs that could still be built, found on the Capacity Remaining page
Hi Ron,
Yeah it’s funny about that, sitting back and having a look there is many things I should of covered and this is something I should have included or mentioned in the book but hey cant remember everything 🙂
The Optimal density for VM:Host is meant to be derived from the average VM specifications.
Unfortunately in my Lab CPU is the limiting factor and both the small and average have basically the same vCPU allocation. So I was unable to really verify within my environment.
And unfortunately this calculation is not something that can be changed at this moment.
hope this helped.
Cheers
Hi Scott,
Is there a way i can change wait or cancel cycle time. i.e can i make wait cycle as 1 minute instead of 5 minutes?
If yes please let me know how to do the same.
Thank you.
Hi Sheela,
Your talking about the wait cycle and cancel cycle when it comes to symptoms yeah?
This is just he data collection interval which can be changed and should affect these cycles
see post https://www.virtualiseme.net.au/index.php/vtips-4-vrops-solution-data-collection-interval/ for changing data collection.
Cheers
Hello Scott,
Many thanks for your reply. I was looking for data collection cycle and i did find the link useful.
Thank you.
HI Scott,
Nice article there… takes away lots of confusion…
How would you calculate capacity. my reports are showing 0 % free resources but i can see atleast 30 % resources available in cluster.
Does it calculate based on consumed memory or active memory & where to set those matrices so that capacity is calculated at actual value .
It will depend on what you have chosen to base capacity on in the policy, in 6.3 onwards you can have Consumed, Allocation and Demand. If you have chosen multiple it will base it on the mose restrictive.
If you browse to the resource and select the Analysis tab and choose Capacity Remaining and expand memory it should show you which one is being use for the calculation.
Cheers
Thanks Scott. .
We have allocation based model in place and on manual calculations found that cluster is 77 % over subscribed which kind of shows why it is crying for capacity .
Was wondering which one will give me better picture.
On a separate topic ,Allocation and demand based are understood . What is the definition of “consumed” in Vrops …I thought allocation & consumed are same or have VMware again changed the definition .
Just wondering what do u means with ” mose restrictive ” . I was thinking of using a combination of demand & Allocation but not sure what would be the result. Will it help me with a more realistic picture.
Also does it help to enable few super metrics in policy . I am using few, which are available on sunny dua,s blog.
Hey Sushil,
Apologies for the delay, Consumed I personally thing will give you the best picture. Now Allocation is how much has been allocated, Demand is how much is needs and consumed is how much it has written data too.
So these days with 64bit Os and ASLR machines will consume close to allocation, but all depends on the operating system and applications.
If you have multiple capacity management calculation selected eg allocation, demand, consumed then the one that shows the lease capacity remaining will be the one vROps will select to have its numbers (time remaining capacity remaining etc) to be based on.
Hope that made sense?
As long as the super metric is something you want to have data on then enable it 🙂
Cheers