--- In ib-support@y..., "Thomas Steinmaurer" <ts@i...> wrote:

> Hi all,
>
> Using EXISTS instead of IN makes the query pretty fast. Fetching

now all rows

> to the client application (to export them into different formats),

the CPU

> utilizitation for both, the client application and the ibserver.exe

process

> is about 45% for each. But after approx. 115.000 fetched rows, the

CPU

> utilizitation for the client application falls to 0% and for the

ibserver.exe

> to 99%. The only way to bring down the ibserver.exe process is to

kill

> the client application.
>
> This happens for different client applications.
>

I have seen such behaviour with interbase when importing a paradox
database on the client to interbase on the server. The problem was
that the transaction grew to big. Committing and opening a new
transaction every 1000 records (486, 24mb server ;-) ) solved the
problem. Have you tried to 'fetch' it in chuncks? It's possible with
the new SQL commands FIRST SKIP.

Peter.