Ann,

The main reason of the cluster thing is for high
availability...something like if the primary machine takes a dump
the warm standby machine can take over. We want as little downtime
as possible, and a failover setup gives us that over a single server
setup.

Past that, I would just like to see the extent and capabilities of
Linux and Firebird, and see if anyone has used it with beasts such
as clustered filesystems like OpenGFS (which might be a nonissue
since it seems the database has to be written to support the kind
of 'parallel' IO operations that a clustered, shared filesystem
uses). Past that, I didn't know if a setup like a database sitting
on a NAS filer and mounted via NFS on multiple clustered machines
running fbserver would have any performance advantages over fbserver
running on a single, fast machine, and if it is doable in the first
place...

From the research I've done already, it looks like we'll be going
with an active-passive clustering setup using drbd and heartbeat.
Both machines will be "high-performance" (SMP and high performance
I/O setups)...

Robby


--- In Firebird-general@yahoogroups.com, "Ann W. Harrison"
<aharrison@i...> wrote:

> Robby,
>
> Out of curiosity, why are you interested in a clustering
> solution rather than a multi-processor w/ high performance
> disks?
>
> Regards,
>
>
> Ann