Hi Jason,
> Sounds greats but one other thing to add... Our back-up servers are not
> on the same network... This is so that if a switch or anythin fails then
> the backups are on a completely septate system if the network settings
> are inherited from the 1st server it will have no connectivity with the
> network.
The clustering isn't aimed at being a backup. It's main goal is to have an
identical second server that's fully up to date and which takes over
automatically and with an almost unnoticeably small delay if the primary
server goes down. Ideally you'd still have traditional backups of some form
or kind (like the built in backup solution) and preferably stored in an
entire different facility (to be prepared against natural desaster, fire,
theft, etc.).
The "RAID over ethernet" is best done with Gigabit links and as few hops as
possible. A direct crossover connection between both nodes is preferred.
Especially as you don't want anyone to sniff that traffic. Encryption or
tunneling it through a VPN would be possible, but requires more CPU power and
lowers throughput, which is not really desired. Unless the benefit of having
it physically spaced outweights the backdraws.
Putting the 2nd node into a different network is an option, but then you'd
need to invest into additional routing equipment. But that's the typical
story: The closer you want to get to 100% redundancy the more expensive the
infrastructure gets. But finally being able to cluster BlueQuartz will be
quite a leap into the right direction. And beyond that? Well, one step at a
time ...
--
With best regards,
Michael Stauber