Hi,
when you spawn the program, how many transactions does it process - sounds
like just one. If so, the spawning will take most of the time. How big are
the 'socket' transactions coming down the pipe - if they are 100 bytes, no
problem, if they are 10,000 bytes, then maybe a problem.

Socket processing is really quick. Why not read from the socket, and write
to a message file. You can then have a process reading the message file, and
processing the transaction.
In fact, the 'reader' could write to several message files, each one being
read by a separate process, and processing the transaction. Again, message
files are really fast.

If you want to discuss this further, off-line, feel free to mail me - pete
at gui-innovations dot com

Cheers

Pete

--
Pete Vickers
Microsoft Windows Embedded MVP
HP Business Partner
http://www.gui-innovations.com

"Jack Nickels" <[log in to unmask]> wrote in message
news:W5KdnYz6co-EmBXdRVn-sQ@fidnet.com...
> Hi,
>
> Having posted this question before, i am looking for some sample programs
in the HP3000. Sorry for the repost :)
>
> We are involved in real time synchronization of data between Oracle and
HP3000 systems and we are working on the various options available for such
a synchronization.
>
> The best method we have hit so far is through NetIPC Socket calls over
TCP.
>
> In this regard, can you please help us on the following.
>
> 1. Pointers in the Internet which says Socket is the best (Performance &
Load) method to communicate between the two systems and why?
>
> 2. Sample programs which enlight how to connection pool in the HP3000
Socket programs.
>
> In this regard, we have built an socket environment in HP3000 where there
is a listener program which is currently monitoring a port in the HP3000 and
this program spawns another program whenever there is a request on the port.
The spawned program reads the data from the port, processes the request,
sends the response and closes the connection.
>
> We are looking at huge number of transactions hitting HP3000 at the rate
of 1000 transactions per second and we are not sure whether the above
architecture will serve the purpose under such a heavy load. (The processing
time for each transaction is within a second)
>
> So we are tying to keep some dedicated spawned processes which will be
running at any point of time. Please help us with some sample HP3000 Cobol
programs if possible.
>
> Thanks for your time and Help
> -Jack
>
>
>
>
> ---------------------------------
> Do you Yahoo!?
> Yahoo! Photos: High-quality 4x6 digital prints for 25
>
> * To join/leave the list, search archives, change list settings, *
> * etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
>

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *