Recently had a good conversation about upgrading from vCOPS 5.8.x to vROPS 6 with Steve Bridle who posted on migration preperation HERE
This got me thinking not only about cleaning up old data but to optimise the data collection. This is something I have not covered before but is something that can really make a difference specially with performance and resource consumption on vROPS.
It is generally a good idea to only collect the metrics we need, and remove the ones we don’t. Most of these we will find will be removed from the Virtual Machine objects, think about it, there is over 2000 metrics collected per Virtual Machine we remove 10 metrics and we have 2000 virtual machines that 20 thousand metrics every 5 minutes not being collected and this is a very small basic example, but will make a difference to both compute and storage required in your environment particularly with larger environments.
Lets now have a look at how we can optimise our metric collection.
With standard licensing we can not change the metric collection other than the 2 choices above. Balanced has about halve the metrics being collected, But I have found metrics that are required that are not in the balanced profile. If we are running advanced or enterprise tho those above profiles can be changed.
2) Here we can see the “all attributes” and “default attributes” for all the object types. The all attribute represent the full profile and the default attributes represent the balanced profile from the admin UI.
3) We notice that these attributes are done per adapter and object type, now select an attribute package and click the little pen icon to edit. We should now see the attribute information. On the left are the metrics we check the checkboxes for the metrics we want to collect and un check if we do not want to collect it. Click OK to save.
4) Ok that’s the standard attributes but we can create our own. To do this select the adapter kind and the resource kind, we can then click the add icon, This will bring up a screen similar to the previous image but no metrics are selected and we have to give it a name. shown in the below image.
Select the adapter kind, here im dealing with the VMware Adapter then select the resource type and on the right there will be a drop down list where we can see the 2 default attribute packages and the one we just created. Select one and click OK to save it
So that’s how we change metric collection for pre 6.0.
The biggest problem here is that this is not very granular object type is as low as we can go.
So the problem I was just talking about is fixed in vROPS 6, Metric collection is now part of the policies, this means we can go right down to a specific object if we wanted. Most the time we will generally only be doing it per cluster but the flexibility is there if we wanted.
2) Now we click on the Override Attributes bar on the left we can see the metrics in the right hand side. We notice down on the bottom right there is 21930 metrics being collected. This will be different depending on the amount of solutions(Adapters) we have installed, either way we do not need all of these.
3) Luckily we don’t have to go through this page by page, we can select a drop down for the object type and can search with the filter text entry. Once we find the metrics we don’t want to collect we change the state to local disable. the opposite is true if we want to collect the metric we select local enable. This is shown in the below image.
That’s it click SAVE and the policy will now collect or not collect metrics at the next collection time.
Stay tuned for more vROPS 6 love.