Booting from SAN has so many benefits, it is sure to become the standard for server engineering. By attaching storage address space to an HBA, the disks can be removed from the internal system and controlled by a tier I storage unit providing many benefits local disk cannot. For instance, should hardware fail on an existing Boot-From-SAN system, moving the disks to a working server is just a matter of re-pointing the WWID to a standby system and booting the address space on the working hardware; no muss, no fuss. From this feature comes one of the kewlest features I employ as often as I can: address space cloning, and thus entire server cloning. By employing storage virtualization (which is a standard feature on pretty much ALL storage devices nowadays...kind of like a CD-ROM or twisty ties) system admins can simply create a clone of storage address space, assign it to an HBA, grip it and rip it, and you are off to the races; a full server provisioning in under 10 minutes...unless you have an hour to waste watching a status bar move from left to right...

The steps are pretty generic amongst all major (and pretty much all minor) vendor storage units:

1) Set you FC-HBA to boot from SAN with a Boot BIOS
2) Attach some storage address space to the FC-HBA's WWID
3) Set the BIOS to boot from the address space assigned
4) Install as normal

To perform the cloning process:

1) Create a snapshot (most vendors have some type of cloning process that does this by default)
2) Clone a new volume from the snapshot
3) Attach cloned volume to the HBA WWID
4) Boot that bad boy

These are pretty high level steps, but really it is as simple as the steps listed above. Of course you have to concern yourself with spindle limitations, bandwidth, and all the other variables required in storage engineering, but is certainly not a process that should be feared because of the stigma on storage virtualization. It is a skill that should be in every worthwhile systems/storage engineer's bag.