ODBC performance problems

[ouuch@xxxxxxxxxxx]

    Hello,

    I have a problem with an ODBC job here, running on an NT server machine.
    It read/writes to a single OS/400 file (via SQL). The software is written
    by a 3rd party programmer, so I can only give tips and hints. And I have
    no or minimal experience with ODBC access.

    On the AS/400, there is an QZDASOINIT job running that handles the ODBC
    requests. I never see any error messages on the AS/400, but at a point in
    time, the job will end and another one start - several hundreds in an
    hour, one after the other. After some time, these submits stop and the
    job runs normally for some hours.
    In NETSTAT (on the AS/400), I also see the NT server opening and closing
    one port after the other - the ports are in status "wait" and will close
    after a time-out, while the server opens new ports.

    So, I'd like to know when OS/400 decides to end/start an QZDASOINIT job
    or which ODBC-Transactions will do this?

    Basically, the NT server regularly checks for new records in the file,
    processes these and will finally update the records.

    The select/updates are working, but only very slowly sometimes.

    I'd like some suggestions what to do on the AS/400 (trace/dump/whatever)
    to get a handle on the problem.

    Thanks for your input,

    Oliver Wenzel