I think we need more info to provide meaningful feedback.

If the bottleneck is really at the hardware level, then you want to go
to at least a 64 bit Opteron box with gigabit SAN connectivity and as
many CPU's as you can profitably use.

If your process is queueable and you are running a multi-CPU box then
running the receiving process on a separate thread from the write
process will allow the queue to buffer spike loads and maximize
throughput

What throughput are you experiencing? What hardware do you currently
have?

On Wed, 2004-11-10 at 22:30, robertgilland wrote:
>
>
> Thanks for your responces.
> But we feel we have already optimised the code to its full
> potential.
> It is at the hardware level where we need to start looking.
> We have a single process injecting and updating data from
> an external source. There are no clients all work
> is done in a single process on the server machine.
> We need to know how we can optimise the hardware with firebird,
> without a trial and error approach which would cost heaps of
> money and time to decipher how to build a machine that can
> take firebird to its maximum potential
>
> Regards,
>
> Robert.
>
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>