Stress measures the amount of workload for a specific object over a period of time. The objects that can be measured include CPU, memory, network I/O, datastore I/O, and vSphere configuration limits. Stress only measures demand, whereas with Capacity Remaining can measure demand and allocation. The default period of time is 30 days, which can be configured in vRealize Operations 6 policy wizard.
Stress is selected by default for Peak Consideration in the Capacity and Time Remaining elements; vRealize Operations uses the stress elements to account for peaks in capacity usage.
The default setting for the stress line is 70%, vRealize Operations Manager calculates the amount of object demand that goes over the stress line during the 30 day time sampling.
Below is the calculation for the measured amount of stress volume:
- Total Stress Zone Volume
- 720 hours (30 days) * (100% total capacity - 70% stress line) = 21,600 Stress Zone volume for the past 30 days
- Measure Object Stress Volume
- Amount of hours object is over the stress line * (100% total capacity - 70% stress line) = Measured Object Stress Volume for the past 30 days
- Measured Object Stress Volume / Total Stress Zone Volume = %Stress
If we use the formula, we get the following results:
- Total Stress Zone Volume
- 720 hours (30 days) * (100% total capacity - 70% stress line) = 21,600 Stress Zone volume for the past 30 days
- Measure Object Stress Volume
- 148.436 hours * (100% total capacity - 70% stress line) = 4,456.08 Measured Object Stress Volume for the past 30 days
- Measured Object Stress Volume 4,456.08 / Total Stress Zone Volume 21,600 = Stress Percentage of 20.63%
Now we turn our attention to Capacity Remaining, we notice that due to capacity restraints with memory I am able to add 0 virtual machines to this host. The Peak Value was 13.08 GB or 126.04% of the Usable Capacity; however the Stress Free Value is 105.86% because it has taken the percentage of stress into consideration. Keep in mind this is my home lab environment; hopefully you are seeing a more modest trend line in your corporate environment.
Usable Capacity * Stress Free percentage = Stress Free resources require for an object, the difference between the Stress Free Value and the Useable Capacity is what is Remaining on the host. Obviously, because of the memory demand, I have 0% remaining. My average demand is 5.22 GB, which is only half of the total capacity, but vRealize Operations is going to take into account the Peak Value and percentage of Stress over the sampling period because it could have application workloads that have high workload periods, such as monthly batch processing or seasonal database transactions.
When we move to Time Remaining, we can see that there are 0 days left due to memory demand and 67 days left due to disk space. Again, it is using the Stress Free Value of 10.99 GB against the Usable Capacity of 10.38 GB.
When using the stress policy in conjunction with Capacity Remaining and Time Remaining, you can identify hosts and virtual machines that require additional resources before you impact business applications. In addition, you could use the VM:Migrate action in vRealize Operations to move virtual machines to an alternate host with more capacity to balance out the environment. This helps you avoid performance problems and helps you get some sleep at night!