I recently migrated some VMs to use Thin Provisioned VMDKs since they had been provisioned with way too much diskspace. Our current setup only uses stand-alone VMFS volumes so it is a manual process to migrate the machines between datastores if and when the datastores are almost full. So I started thinking about datastore clusters (which I think are a pretty cool thing). How would this be managed in datastore clusters ?
After spending large amount of time googling and reading up on vSphere documentation I was still not 100% sure on how this worked exactly. So I sent a message to Larus Hjartarson (@lhjartarson).
Of course he had the answer 🙂
So – I decided to write this here so I would at least remember this:
StorageDRS will monitor the disk space. You choose the high-watermark for the datastore cluster. You also choose the watermark for how much would be gained from moving a VMDK from datastore X to Y.
StorageDRS will then monitor the datastore cluster every 8 hours (or any given time you configured). If it finds that datastore X or Y is nearing high-watermark and gain X percentage by migrating VMDKs to another datastore in the cluster it will either notify you or take action by migrating the VMDKs (by default it only notifies you).
Now, let’s go implement datastore clusters !