Best Practices to Minimize Downtime During Azure VM SKU Upgrade

Question

Dear All,

I’m looking for best practices to minimize downtime while increasing the SKU size of our existing VMs running in Azure. Could you please provide the recommended approach for this task?

We have 4 servers (2 x A and 2 x B) configured in an availability set. The load balancer's backend pool has 2 NICs, each with one A and B servers.

The requirement is to increase the SKU from F series with 2 vCPUs to 4 vCPUs. What is the recommended approach to complete this process?

Answer 1

Hi,

You can minimize downtime by first setting health probe to fail on VM you want to update, next give it time for any existing requests to complete, resize VM, when VM comes back up after resize set health probe back to success, repeat with next VM. In this way you are only updating one VM at a time and leaving the others operational.

When I say "set health probe to fail on VM" I mean configure/modify that VM so that health probe will fail. For example, say you have health probe requestPath set to /healthcheck.aspx and this has script that returns HTTP status 200. If you temporarily rename healthcheck.aspx to healthcheck bak.aspx on the VM you want to update it will cause the probe to fail.

https://learn.microsoft.com/en-us/azure/load-balancer/load-balancer-custom-probe-overview

Please click Accept Answer and upvote if the above was helpful. If something is unclear please add a comment below.

Thanks.

-TP