At 02:06 PM 10/08/2005 -0600, you wrote:

>Maybe set the affinity to use just one of the CPU's for FB server,
>or does that apply to windows only ?

It doesn't apply to Classic at all, on Windows or Linux.

>We have a Xenon 2 CPU Fujitus Siemens RX200 S2 with 4 GB ram and win
>2000 SP4 on it.
>The swapfile is 2gb - 4 gb
>We are running a Firebird Classic Database.


>Our problem is that the Firebird process ist always swapping and we
>have very much page faults. The E/A Read ist at 4 TB in 2 weeks.

What is the "E/A Read"?

>The database is 800 MB large
>Who has some ideas what we can do?

The first thing to look at is the cache size. If the database cache is
thrashing the swap file on a Classic server, it's often a sign that the
static caches, that are allocated for each connection, are too large for
the RAM available on the system. Multiply the cache size by the number of
currently-connected users to work out how much RAM is being statically
allocated to page caches.

4 Gb seems like plenty of RAM, but it would be of interest to try and
estimate how it is being used and is getting exhausted often enough to
cause this kind of disk-thrashing.

Are you running other apps on the server, besides the Firebird server?

Do you have multiple hard disks on the host machine? Do you have temp
files specifically allocated for sorting? If so, are these files located
on the same hard disk as your swapfile?

./heLen