Kenneth Burgener wrote:
My second question is, if I did do the RAID 0 setup, and if I noticed
one disk begins to fail, is it possible to somehow transfer the
remaining good portions to a different disk without having to rebuild
the whole array? The problem is that I am planing on having 4 x 250GB
drives. At 1 TB there would be no place to move ALL files off so I
could rebuild the array.
This setup is for a multimedia setup, so I am not too worried about
loosing some files, but I would like to be able to salvage what I could.
As others suggested, LVM can probably give you the best use of your
space. If you simply laid out a single filesystem across all 4 drives,
you'd have to choose software RAID 5, because RAID 0 across 4 drives
won't last long. With RAID 5, you'd give up 25% of your space and you'd
lose everything if you lost any two disks at once.
With LVM, however, you can create partitions only as needed. Create a
RAID 0 array for daily work, since RAID 0 across 4 drives really is
quite fast, and you can probably bear the loss of a few days' worth of
work. The same set of drives can also hold RAID 5 arrays for archived
projects and RAID 6 arrays for important archived material that needs to
survive even if 2 drives fail.
However, that's all theory. The truth is that you should be worried
more about user error than hardware failure. I've been watching the LVM
mailing lists, and there appear to be several ways to destroy LVM just
by typing the wrong thing. This is why RAID is no substitute for
backups. I've been using rdiff-backup[1] lately and I really like it a
lot. It's like rsync, but it produces incremental backups. I set it up
to run as a daily cron job.
Shane
[1] http://www.nongnu.org/rdiff-backup/
/*
PLUG: http://plug.org, #utah on irc.freenode.net
Unsubscribe: http://plug.org/mailman/options/plug
Don't fear the penguin.
*/
