Poop, sorry about the empty reply.


...

> > Most likely the above is an over-simplistic answer, but all
> > I've ever used over the last four years is classic, and no
> > query/export/report has ever blocked other people from
> > working...
>
> How many users have you managed to support with Classic? (How
> much memory did you throw at it to manage it?)

We pool connections for about 200 users, so there's only about 10-15 active
fb_inet_servers going at any one time - not a large system by any means.
4-6pm in the evening gets busy, with the load average reaching 6's though!

> The problem I was discussing is not the old SMP CPU thrashing
> problem. The pausing issue is something quite different.
> There is no "blocking", indeed activity on other connections
> is desired because this wakes up the transfer process.

I realise it's not the shuffle bug, but you're still using an engine that
(largely) serializes operations on a machine that CAN run things in
parallel...

Forgive me if components of this reply are incorrect - I'm still learning...

Nige.