In an animated thread inside VMTN Forums just appeared a neat witness of how important and how reliable VMware VMotion is.
In my humble opinion it’s worth more than a thousand of case studies.
I quote it integrally:
My Virtual Center console shows 215 VMs provisioned, 806 VMs migrated.
I move live production servers around ALL the time via VMotion and have never had a problem.
And I work at H&R Block, and it’s tax season. Believe me, I DON’T want servers going down this time of the year.
I had a memory module show as degraded the other day. The server didn’t go down, but Insight Manager was telling me to replace it before it failed. I moved all the VMs off in the middle of day and replaced the memory, and then moved them all back.
This server was hosting 18 VMs. I don’t even want to think about hard it would have been to find a common maintenance window for 18 different servers in order to take them all down at once.
I’m sure I would have had to drive 45 minutes into work at 3:00 am, replace ONE memory module, and then drive 45 minutes back home.
Instead I VMotioned 18 VMs off (2 or 3 to each of my other servers), replaced the memory module, and VMotioned 18 VMs back. While eating my lunch.
The trick is design your capacity to give yourself the ability to recover from a completely failed VMware server or in case of maintenance work like this.