> I am in the process of selecting hardware for a 7x24, mission
> critical application (I hope you like jargon). I am evaluating
> A Tandem Fault-Tolerant UNIX-based system, a pair of Data General
> Aviions, and a pair of DEC Alphas. I would like any feedback about
> the reliability of any of these systems. We will be running oracle7
> in a realtime enviroment. Basically, I can not afford for the system
> to be down for more then 3 minutes at any given time. I am
> expecially interested in horror stories.
Not the usual sort of fault-tolerant system, and not as bulletproof a
solution as a Tandem or Stratus, but probably cheaper. I configured a
system like this as the primary output spooler for a newspaper; they
can afford small amounts of downtime, but need to be back up in a very
few minutes. They have two identically configured Sparc 20s with two
Fast SCSI/Buffered Ethernet (FSBE) cards and two external 1MB drives.
All the important data (including the root partition) is on the
external drives; each drive is connected to its own FSBE, and the
drives are mirrored using Sun's Online:DiskSuite software. If a drive
fails, the mirror will continue running with no downtime. Likewise for
one of the FSBE's. If the system itself fails, the monitor, keyboard
and disks get plugged into the second system, which boots while they
place a service call.
We can shout "Fire!" in a crowded Central America.
- Ian Shoales