How is the VM availability metric calculated ?

Question

Can someone explain specifically how the availability metric is calculated ? I see at the bottom of this page - https://learn.microsoft.com/en-us/azure/service-health/resource-health-faq?wt.mc_id=knwlserapi_inproduct_azportal#why-am-i-seeing-the-vm-availability-metric-reflect-vm-downtime-when-i-observe-no-impact it states "The VM availability metric is calculated as an aggregation of the health checks stated here." however it also states "We have plans to move away from the aggregation model and rely only on one, unified high confidence signal from the host server. This work is already underway and will ensure our monitoring offerings give consistent values. We anticipate this work to be completed through the course of 2023."

Can someone confirm if these are the checks that are still used to calculate the availability metric or if it has changed to something else?

• Is the server hosting this virtual machine up and running?
• Is the virtual machine container provisioned and powered up?
• Is there network connectivity between the host and the storage account?
• Is there ongoing planned maintenance?
• Are there heartbeats between Guest and host agent (if Guest extension is installed)?

Answer 1

Hello, @Nick Olano !

How is the VM availability metric calculated?

The work to move away from the aggregation model and rely only on one, unified high confidence signal from the host server in 2023 is now complete in the form of the VM availability metric. I reached out to the Virtual Machine team and confirmed that the exact workings of the VM availability metric as well as general descriptions of the checks that make up the VM availability metric are not publicly available.

What we can say officially is that "VM availability is computed based on different signals from the host."

The values of the VM availability metric and their descriptions can be found here:

https://learn.microsoft.com/en-us/azure/virtual-machines/monitor-vm-reference#vm-availability-metric-preview

Value	Description
1	VM is running and available.
0	VM is unavailable. The VM could be stopped or rebooting. If you shut down a VM from within the VM, it emits this value.
Null (dashed line)	State of the VM is unknown. If you stop a VM from the Azure portal, CLI, or PowerShell, it immediately stops emitting the availability metric, and you see null values.

---

While I realize that you were looking for a very detailed breakdown of the VM availability metric, the I hope this has been helpful nonetheless. Your feedback is important so please take a moment to accept answers.

If you still have questions, please let us know what is needed in the comments so the question can be answered. Thank you for helping to improve Microsoft Q&A!