After my first post on the vCops improved Capacity Remaining Badge, my former colleague and friend Bouke Groenescheij posted an interesting comment: “May I also recommend changing the ‘Capacity Remaining is based on:‘ to Usable Capacity. This will take HA and overhead in consideration. This will result in a lower amount of deploy-able VMs, however, when facing a host failure you know for sure you’ll have sufficient resources.”
Well that’s certainly a good tip, so…let’s take a look at this option and also evaluate the optimized ‘view’ for Capacity Risk.
About Usable Capacity
Let’s take a look at the “Change Default Policy Configuration” and select option 3a Capacity and Time Remaining. This screen let us choose Capacity Remaining to use the Physical Capacity or Usable Capacity. When choosing the latter option, vCops will take in account HA settings: “Usable capacity includes physical capacity minus buffers and HA policy. See 3b Usable Capacity for settings”.
After choosing “usable capacity”, open option 3b: Usable Capacity. Let’s have a closer look at these settings:
You can set Usable Capacity Rules and you can set Capacity Calculation Rules.
The Usable Capacity Rules let you specify a spare capacity percentage, which should be reserved as a buffer. Buffer resources are not used when capacity remaining is calculated. Important; when selecting Use High Availability (HA) configuration, and reduce capacity and when HA is actually configured on your cluster, the buffer settings for CPU and Memory are ignored. vCops will use HA settings in this case.
Regarding the Capacity Calculation Rules VMware is telling us:
- Use last known capacity- If the interval included in the view is two months, and today is June 30, views
of capacity such as Average Virtual Machine Capacity show you the capacity
used on June 30.
- Use actual capacity – If the interval included in the view is two months, and today is June 30, views
of capacity show you the average of capacity used in May and June.
Capacity Remaining – Recalculated
Adjusting these settings will result in a recalculation of the remaining resources in your cluster. In this example I’ve set a buffer of 50%, this results in the following Virtual Machine Capacity calculation:
In this example the calculation is independent of actual HA settings and solely based on the 50% buffer settings. The calculation is telling us we’re over committing the cluster with 3 virtual machines, taking in account a 50% buffer setting (which can be an actual setting in a 2 host cluster).
The warning/over commitment is also displayed in the Capacity Risks Details view:
I hope this extra additional article on the Capacity Remaining Feauture gave you some additional insights on how vCops calculates the remaining capacity. Be sure to also read part 1 of this article! Also check Bouke’s postings on vCenter Operations Manager on jume.nl.