For a couple of months now, I’ve been hearing about the upcoming 4.0 firmware and Auto-Snapshot Manager, VMware edition for the Equallogic PS series SAN. This new snapshot provider would allow us to coordinate snapshots between VitualCenter and the SAN, and, according to Dell/Equallogic, allow easy restoration of a single Virtual Machine from the SAN-based snapshots.

I had the opportunity on Thursday night to watch a pre-recorded demo as well as to attend a live webinar on Friday morning. I must say that after these demos and seeing exactly what this product does, I am stuck somewhere between excitement and disapointment. It’s a really cool concept, but I believe it still needs a lot of polishing, especially on the recovery side of things.

There are lots of awesome features in the new snapshot provider. We will have the ability to automatically, in a single click or scheduled task, trigger an ESX snapshot, including memory dump, then snapshot the SAN volume, followed by removing the ESX snapshot. This eliminates the journaling effect and associated performance hit and disk requirements of the ESX snapshots. This is all handled through a nice web interface, and the VirtualCenter folder tree is carried over, allowing snapshot schedules to be applied to groups of Virtual Machines.

There are, however, some catches. Only the selected VM’s are triggered for ESX snapshots, but the entire SAN volume, which may contain many other VM’s is snapshotted. This makes the ability to group VM’s using VirtualCenter folders less than useful. Let’s say I have four VM’s split between four volumes and want one machine on each volume to be snapshotted everu 12 hours. Then, I want one machine per volume to be snapshotted every 24 hours. In this scenario, I will actually end up with, at the SAN level, two snapshots per day of both entire volumes and all the VM’s since the entire volume is snapshotted. So, in my opinion, snapshotting VM’s by any grouping other than an entire SAN volume isn’t going to be practical without a lot of wasted disk space.

On the recovery side, I think there is a lot of room for improvement. It is very easy to revert an entire volume and all the VM’s it contains. Beyond that, restoring a single VM, for example, becomes a somewhat lengthy process. Basically, it involved going back to the Equallogic Group Manager, setting the snapshot online, going to ESX and mounting the snapshot as a new volume, deleting the damaged VM, copying it manually from the snapshot to the production volume, adding it to inventory, booting it up, and then unmounting the snapshot. Alternately, the VM can be booted from the snapshot volume, then migrated back to the production volume using Storage VMotion. Storage VMotion, however, requires accessing the ESX command line.

It is my hope that, in a future release, Dell will automate some of the recovery process using the VMware API’s. Currently, there are lots of improvements in creating the snapshot, but no real change in the process of recovering a VM.

I am looking forward to getting the Auto-Snapshot Manager, VMware edition installed in our environment and actually seeing it in action in a production environment. Expect another post in the future with more details once I actually get this up and running.