2009-03-24 01:32 AM
If we roll out all our VDI desktops using FlexClones and RCU v2, how do we put together a plan for replication and failover? Do all these VDI desktops need to get recreated? Can RCU be fed a policy or scripted so that when DR is invoked, it can get immediately into the process of re-creating all the VDI desktops, or is there a better way?
Solved! SEE THE SOLUTION
2009-03-24 05:45 AM
Are you most concerned with loading the VMs into Virtual Center and your connection broker? When RCU v2 creates the FlexClones they are fully created VMs and have no dependancies on each other. Therefore if you SnapMirror the volume to a remote controller all the VMs in the volume would be replicated as well. Since SnapMirror keeps data in it's deduped form this will actually be a rather easy replication with not much going over the wire.
I do see your point though since at the DR site you now have to load the VMs into Virtual Center and your connection broker. The loading into Virtual Center isn't that hard and is just a little bit of shell scripting. Loading into your connection broker will depend on which broker you have and how they load target VMs.
I hope I didn't miss the mark on what you were asking here, please let me know if I did.
2009-03-24 05:56 AM
So how does this tie in with FlexClones? If I have deployed all my VDI desktops from a base image, or a group of images, these will be using very little storage on the primary. But to SnapMirror these, would this not break the FlexClone relationship to the base image at the destination? Is there any ability to replicate FlexClones without the overhead?
2009-03-24 06:08 AM
RCU uses the new File Level FlexClone for its clones. This is important as FLFC works differently than the traditional Volume Level FlexClone you might be used to. With FLFC we do create a backup snapshot, then clone the meta data required for cloning the VM, then remove the backing snapshot. That last bit is the important bit. This means that after the FLFC completes that there is no dependancy on the base image at all. It really is just as if you populated the volume with full VMs then deduped it back. Same end state.
If you create a new volume of VMs based on a VM image the tool creates a new volume, then copies the first VM into the volume. Once the copy completes then it begins the process of creating FLFC within the new volume. This process removes any dependancies on the base images or between volumes. When you are replicating you don't have any overhead of the FlexClones and you are right, very little data will be transfered.
2009-03-24 01:45 PM
Sorry, there's me getting Volume FlexClone and File Level FlexClone confused! Makes perfect sense what you point out and obviously there would be no knock-on effect with SnapMirror.
It will be interesting to see how things develop with this and if there is any future plugins or integration with Site Recovery Manager.
Cheers for the input!