Subject: Re: differential scsi controllers
To: Herb Peyerl <hpeyerl@beer.org>
From: Jordan K. Hubbard <jkh@time.cdrom.com>
List: current-users
Date: 07/02/1998 05:42:19
> define "fragile as hell"?  Sure, there may be some missing features
> but as far as "large fast striped volumes", I've _never_ had a problem
> with any of my ccd's except in the case where one of the disks failed.

That latter case where the disk fails is precisely what makes it
fragile, and not because recovery from hardware failure is something I
expect in any software feature set.  No matter how you slice it,
having your point of failure for a single volume multiplied by n
(where n is the number of physical disks involved) is just bad news.

Disks, and the kinds of disk-support environments one tends to find in
most PC cases today, are not yet reliable enough to where I can take
(for example) 5 drives and ccd them together without *substantially*
increasing the chance of catastrophic failure for whatever data is on
that volume.  That can certainly be mitigated by making regular
backups, keeping redundant copies of the data, whatever, but I still
don't like the numbers I get when I take the TTBF for a single drive
and divide it by 5 to estimate that ccd's chances of providing
uninterrupted service for 12 months.  I would personally call those
numbers evidence of "fragility" but, of course, YVMV (your vocabulary
may vary :-).

I'm also not saying that I hate ccd's and wouldn't recommend one for
anything since that's not true.  I recommend them for any situation
where you have lots of data that can either be regenerated at will or
just isn't too terribly important (packet sniffer files, non-vital log
information, USENET news, porn GIF collections, etc. :-).  If someone
also wanted to add RAID>0 support for ccd then that'd be wonderful, of
course, but I'm not exactly going to hold my breath for that (or
http://www.lemis.com/vinum.html for that matter).

- Jordan