Not entirely true. If the cache hasn't dumped changes to the files
prior to the crash, they will be lost when the spare kicks in. Further,
these setups (the ones that I've seen) don't really require any
"special" hardware. You just put both machines on a single SCSI chain
and make certain that you don't access the disks from the spare unless
the original is offline. I know you have to be certain that the two
controllers have different SCSI IDs, but other than that, I'm sure there
are other details to take into consideration. Of course, I can think of
an easier solution if you have time to let the spare machine boot before
it kicks in...
A software solution like you describe that is updated real-time would
have a latency problem similar to a VERY exaggerated version of the
cache problem on the hardware solution. The primary machine could crash
while sending updated data to the secondary, in which case that update
would be lost. If you wait for the replication to take place and run
your local filesystems in-sync with a network replication, you're
certainly going to slow the filesystem access speed to the point where
it would be unusable (unless you're going to sustain 60% theoretical on
GBE or something comparable for your drive replication, which is cutting
into the cost efficiency of an all-software solution at that point).
Just my half-nybble.
--
Jeremy Stanley Trend CMHS
Network Engineer http://www.trendcmhs.org
The opinions expressed herein do not necessarily
represent those of Trend CMHS or Trend Foundation.
"I program my homecomputer; beam myself into
the future." --Kraftwerk, 1981
> ----------
> From: Tom Kunz[SMTP:[EMAIL PROTECTED]]
> Sent: Monday, October 04, 1999 6:11 PM
> To: [EMAIL PROTECTED]
> Cc: [EMAIL PROTECTED]
> Subject: Re: networked RAID-1
>
[clip]
> mount the array simultaneously. Any node can go down at any time,
> regardless of any cron schedule, and no data will be lost. But thanks
>
[snip]
