Helen,
sorry for the delay,

> Have you checked to see whether the IB 6 server is running with Forced
> Writes false (Asynchronous Writes true)? (IB 6 installs itself with

Forced

> Writes false, by default...) If so, and your application is allowing users
> to hold transactions open for lengthy periods and/or you have clients

doing

> large batches of DML, then you could stack up a lot of "unfinished
> business" in RAM. It's just a thought...

Presently i'm not using Forced Writes. So i'm starting checking my apps to
understand how many DML it produces without committing.

> >and i'm testing Firebird 1.0 for the migration.
>
> Firebird 1 should have installed itself with Forced Writes true. How hard
> would it be to run it with similar load and see if you notice a difference
> in RAM consumption?

I'm presently testing it and i'll give you an answer ASAP

> >The databases engine (IB6) is installed on a RedHat 7.1
> >server machine.
> >When the engine start it takes about 18 Mbyte and 4 processes. Then i

launch

> >some application that use 2 different mono-file databases of maximum 10
> >Mbyte size. With about 3 simultaneous connection on the first database

and 5

> >on the second one my interbase size reach after some days about 120 Mbyte

of

> >memory with about 8 processes and keep on increasing.
>
> Is it Classic or Superserver? Classic allocates a pre-defined chunk of
> cache memory for each independent connection process, whereas SS allocates
> one chunk which it shares with all of the child processes which are

created

> for each connection.

Presently is a SuperServer installation and with a maximum amount of 10
simultaneous connection i don't know if it's a good choice.

> >In order to understand if it's a bug in application or an IB normal
> >behaviour i need to make me an idea on how to evaluate the maximum size
> >that such database might reach in memory related to the maximum number of
> >active connections,the database file size and the IB sever configuration
> >params.
> >Anyone can help me to find out a method or an explaining document to

solve

> >this memory evaluation problem ?
>
> There are algorithms around - you might care to search the mers.com
> archives on 'cache', especially postings by Ann Harrison and Bill
> Karwin. Google search may yield something useful, also.
>
> You might find some the following helpful:
> http://www.ibphoenix.com/main.nfs?a=ibphoenix&page=ibp_config
>
> Older:
> http://www.ibphoenix.com/main.nfs?a=ibphoenix&page=ibp_conf_opt
>
> heLen

Thank you is what i'm looking for.

Alessandro