Hi again graf.
Maybe this example will explain better why you do your backups more by filesystem than disk.
Let's say you have 6 disks. Two of them belong to vg00. That is where your operating system is and a make_recovery tape will back that up. (You can also do regular backups of the filesystems in vg00 through sam, too.) The make_recovery will configure the disks for vg00 during the recovery. The HP system will recognize the tape as having the operating system on it, even though there is no operating system on a hard disk.
The other 4 disks are in vg01. The disks are named c1t0d0, c1t2d0, c2t0d0, and c2t1d0. Each of the disks is 9 GB.
In your volume group vg01, you have 2 logical volumes, one named lvbig and one named lvlittle.
lvlittle is 2 GB and its filesystem is /little. It is on c1t0d0.
lvbig is 12 GB and its filesystem is /big. 7 GB is on c1t0d0 and 5 GB is on c2t0d0.
Let's say c2t0d0 fails, and your logical volumes are not mirrored, so you have to recover the data that is on that disk.
Once the disk is replaced and configured, you pull out your backups of /big. You will have to recover the whole 12 GB, not just the 5 GB that were on c2t0d0.
And that is why you back up filesystems, not just disks.
You can find out which disks your logical volumes are on by using the lvdisplay -v <LV path, such as /dev/vg01/lvlittle> command.
Pipe the output to more. After about 15 lines of information about the logical volume, there is a section called Distribution of logical volume. The names of the physical volumes the logical volume is on are listed.
And a final thing: I have to admit I'm not sure what the operating system will do when you replace the disk. You may have to do some configuration work (bring the disk into the volume group, and then rebuild the logical volumes and filesystems).