At 04:49 PM 12/20/2004, Rick DeBay wrote:

> > Doesn't it seem simpler just to eliminate the need for connection
>pooling with cheap connections?
>
>As long as you can have cheap connections, and we don't lose statement
>pooling in Jaybird (which I think still offers a scalability boost).

One of the goals of Vulcan is to pave the way to statement pooling.
We actually considered statement pooling when designing the architecture
in the early dark ages and rejected the idea as using too much memory.
Happily, memory is somewhat more available now than it was in the early
1980's - well computer memory is, mine's pretty much shot - and one
reason for moving the DSQL processing into the engine is to be able
to maintain a cache of compiled statements for reuse.

Regards,


Ann