VROPS 6 oversized VM report

[primetimerob]

I just upgraded a few weeks ago from VCOP 5.7 to Vrops 6.0.1 

In VCOPs we had over 400 Vms listed as "oversized".  In the new Vrops with the same data migrated, we show only 30?  That is quite a difference.  Even if the analytics was tweaked, I don't see how it could say we only have 30 whole VMs in our entire environment that are "oversized".   Anyone else run into this, or have any ideas on the drastic change?

[MacVay · Accepted solution]

It all depends on your Policy....

The default policy in vROps will only list oeversized if the recommended saving is above 50%.

Example:

If you had provisioned 10 GB of RAM the recommended would need to be 5GB or less for it to show in the report.

The default policy in 5.8 was brutal with it's recomendations as typically without adjustment you would end up chasing your tail with right sizing.

The vROps policy is much less agressive and realistic.

Cheers,

[primetimerob]

So if it has to be over 50%, for VCPUs, that are configured at 4, it would have to think you could drop it to 2 or less, for it to show up?  Going to 3 would be 50%, which is not above that threshold.

[MacVay]

Yes,

You can edit this policy to be more aggressive if you wish.  It would be under the analysis section of the object type for virtual machine.

Cheers,

[primetimerob]

Thanks, I've found that policy and I'm seeing what results I get when I tweak it. Thanks.

[dsohayda]

I ran into the same problem despite migrating my data from vcops.

Do you have details on what settings in the policy need to be modified for this report?

[primetimerob]

You go to Administration->Policies->Policy Library

Edit the policy you have the setting

Mine was set in Default Policy, it depends on where yours is set.

Edit Selected Policy

vRealize Operations Adapter - vRealize Operations Analytics

Reclaimable Capacity

Flag as Oversized When.

I've adjusted mine, and haven't seen any difference yet. I'm not sure if it takes time to take effect.

[dsohayda]

ok, thanks. I had the right spot, but was looking at the wrong object type. I had filtered the vcenter adapter cluster compute resource in the override analysis settings of the default policy.

I adjusted the vrops analytics reclaimable capacity to be flagged as oversized when the recommended size is less than 20% of the current size rather than the default 50%.