DSXchange

Funnel stage to bring the streams together and Aggregator stage to do the total. Don't be scared off - the Aggregator stage in parallel jobs is much more robust than its server job counterpart.

Have you validated the link names across the shared container's boundaries? You will need a server Transformer stage within the server shared container to perform the actual lookup - a server shared container containing just a Hashed File stage will not work properly in a parallel job. You probably ...

Put cat as the command and #directory#/test as the command line arguments.

Do you have the Orchestrate manuals? They are available for the asking through your support provider.

That part is not.

Currently IBM is hoping for second quarter 2006.

That's not really how parallel jobs work. How would you guarantee that your routine is executing in the same partition as the lookup key and the reference key? You should strive for an approach that uses a Lookup stage, maybe feeding the overall result into your routine.

rcil wrote:Ray,

Thank you for all the help. Your answers saved me many times. I am just curious how I see your 10000th post?

Thanks,
rcil

Don't know. Did about 20 this morning - it was among that lot - wasn't really paying attention.

Do we need a separate forum for Douglas Adams quotes? They seem to be proliferating in signatures.

You could ask one of the professionals here to develop it, in exchange for money - probably for less than IBM would ask.

Another nice tool in Hawk is an estimator. :D
But only for parallel jobs.

The correct estimate for disk space: "more".

Muriel, this routine is not part of DataStage. You can download it from Kim Duke's tips page.

Here's a great opportunity to use BCI functions. The environments for localuv are already set up, available through variables @HENV, @HDBC and @HSTMT. So all you need is one call to SQLExecDirect(), two calls to SQLBindCol() and one call to SQLFetch() (plus appropriate error handling) and you are th...

Welcome aboard! :D

Thank you for posting a solution, and congratulations on the fact that your first post is a solution!

In your job activities, specify "reset if required, then run" as the run mode. Skipping is managed via the compilation options in the job properties window of the job sequence itself; the check box names are quite self-explanatory, and there's always Help.