Hi Claudia,
In-place hypervisor conversion does seem like a way to go. Please see Prism Web Console Guide for the details on the process.
There are a few different methods that you can go with on this. What I personally found that works with least amount of downtime is to do something similar to the alternative method:
- Move all of your VM’s to 1 cluster (assuming it has capacity)
- Run the in-place conversion (Convert Cluster) utility in Prism to switch it to AHV. Trust me, it works well!
- Use “Nutanix Move” to move your VM's from the ESX cluster to the AHV cluster. It’s a little tedious, but it can handle the driver installations for you and you can reboot the VM on your schedule when you cutover.
- After all VM’s are moved over, convert the other cluster to AHV.
This process gives you a solid rollback option to VMware if AHV doesn’t work out for you. Make sure your VM’s can support AHV. If you have any OVA’s, check for a KVM-version of the appliance.
I know there are many different ways that you can do these conversions. Your account team would probably be able to help you go through all of the details within your environment. I am doing a conversion that is a variant to what I wrote above and it works well.
The best advice I can give you is to know your environment and what outage windows you can have. This is not a zero-downtime conversion. You will need to take a reboot of every VM that you have. You’re changing the hypervisor definition, which the OS sees as a motherboard replacement. It’s magical, yet, when you think of it in that context, you can really see the big picture.
Good luck! Hope it goes well!