Re: Whither RAID?

From: Jack Peacock (peacock_at_simconv.com)
Date: 08/26/04 

Date: Wed, 25 Aug 2004 17:24:07 -0700

"Dr. Dweeb" <dr@dweeb.com> wrote in message
news:cgiau4$1ump$1@news.cybercity.dk...
> It is a common mistake to think that RAID-X is some kind of panacea. Only
a
> correct and regular backup strategy can save you from the sort of failure
I
> experienced. Indeed, it may be exactly this type of failure that the
> article referred to.
>
As I tell customers who think RAID equates to no tape drive needed, any
newbie with the admin password can go and delete system files. The RAID
system will faithfully delete the files across all the mirrors, so that all
disks in the array are equally corrupted.

The problem sounds like a restore to the last good backup, which may have
been a few days back before the file corruptions started. Even a backup is
no guarantee if the problem isn't caught until days later. I had that
problem on an Alpha 2100 a few years back. An pattern sensitive
intermittent failure developed in the data bus to the SCSI controller.
Every few hours a disk or tape write would be corrupted, but in such a way
that it veriefied on the backup. Since the apps were 99% read, 1% write the
problem didn't reach crash proportions until more than 4 weeks after it
began. At that point the entire monthly cycle (31 tapes) had some level of
corruption. I had to restore data files as best I could, patch the sections
back together, and in some cases data had to be reconstructed. The saving
grace was that they didn't work on weekends, so some of the tapes were more
than 31 days old. A nasty hardware failure as it didn't log as a error with
the motherboard SCSI controller. It was a very long week.
  Jack Pea***