One of the things I’ve liked about it is being able to throw new software at the system with knowledge that the OS drive is protected by a snapshot as well as the data volumes.

(Windows supposedly does this through it’s own system restore setup, but i’ve never seen it work completely because it’s not a block-level copy, and tends to leave remnants of the changes behind in the form of DLL files that for one reason or another don’t get deleted.)

In practice, the same simple underlying hardware swap is not always as easy on AIX or HP-UX (I’m uncertain on Solaris). Not going to a new generation of hardware, anyway. I’m not saying it couldn’t be done.

However, even if you exclude those from the list, the problem that I have with actually recommending boot-from-SAN in the datacenter is that it has repeatedly proven difficult to educate the server admins on how to configure it.

That and typically by the time a customer decides to do a hardware refresh, it’s probably a good time to talk about a new OS install anyway (either because it’s a newer OS, or just to clean out the cobwebs on the old one).

That being said, boot from SAN does offer lots of benefits, particularly the data protection options that you mention.

I think in a well-maintained environment where you have a great technical team (you know, that pie-in-the-sky perfect place to work that I’ve never seen), then it can be used to great effect.

I’m actually looking forward to a day when we can boot from a write-protected OS image. All that would need to be read-write would be swap space and maybe a handful of local configuration files. The management and security advantages to such a setup would be fantastic.