autofailover: All or nothing?

[webguyz]

I currently have a dss v6 with File i/o volumes doing replication. If I want to do autofailover I need to create Block I/O volumes. Can I create new Block volumes and then set up autofailover and then slowly copy my hyper-v vhds onto the new block I/O volumes that I'm going to set up with autofailover? I want my File I/O volumes still accessible and not participating in the autofailover setup. Trying not to disrupt my production systems while setting up autofailover.

I'm thinking I probably need to add a few network cards in my primary DSS that will not participate in the autofailover (with virtual Nic) and connect my servers using these temporary NICS to connect to my existing File I/O volumes. Then proceed to set up autofailover creating the new Block I/O volumes and my existing 10g NICS with a virtual nic. I should be able to test that this is working by binding the new virtual nics in addition to the working IP's (on temporary nics) for my File I/O volumes to my servers and accessing both the file I/O volumes and the autofailover block I/O volumes from the servers. Once everything is working I can start copying my data from the File I/O volumes to the Block I/O volumes with failover.

Is my understanding of this process correct? Can I do this while keeping my production systems up? Can I mix autofailover volumes and non-autofailover volumes on primary DSS?

Thanks!

[webguyz]

Has anyone gone from a single DSS V6 to autofailover with 2 while preserving their production data and minimimizing downtime? Or has everyone just created 2 new servers and then moved their production data after the autofailover is in place?

[enealDC]

To specifically answer the q, it's not all or nothing. It's certainly easier to start off with 2 nodes, but you can certainly migrate to a HA setup.
If you are 100% you are moving forward, buy the other node and make sure you have at least 6 - 8 NICs. Once you have the nodes, it's really just playing musical chairs. Get the node setup. Create your targets, move your VMs over one by one until they are completely off of the original node, configure failover. Once your VIPs come online, add the portal to the initiator on a single server then migrate all the VMs onto that single server. Repeat for all servers minus the migration..