DSXchange

Once you know the value of nnn, of course.

You need a minimum of two jobs and a job sequence; one job to get the job number, the job sequence to generate the log table name and pass it to the second job to extract the log information.

If there is, it will appear in the drop down list. (I do not have access to parallel jobs at the moment to check, so you'll have to check.)

If it's really generic you could hard code it. You could introduce it via a job parameter (maybe the $USER environment variable).

Sort with a Sort stage and have it generate a key change column. In a downstream Transformer stage reset the counter to 1 if the key change column is 1, otherwise increment the counter.

It's the restart logic code. That's why I suggested recompiling.

Losing 90% of the columns would be favourite.

They use memory while they can (perhaps up to some configurable limit, as in the Sort stage) then spill to disk.

Beyond that it's a factor of the data volumes. There are no "general" answers. If the whole thing can be done in memory it will be.

Configure the maximum possible (in the data) number of columns and allow them to be NULL if not populated. Run these rows through a Pivot stage.

Forget rows per second. The seconds includes the time spent waiting for the Oracle query to deliver the first row, as well as the time taken to process all rows. Further it will vary depending on row size and a number of other factors. Total execution time should be your metric. The job log (the tim...

If you execute this code in another Oracle client (replacing the parameter marker with a value, the same value that will be supplied by the Key column in DataStage), do you get the same error?

For preference, DataStage BASIC code should be compiled from within the DataStage environment, using the BASIC verb in VOC. As daignault points out, this invokes the nbasic executable (not the basic executable). Nobody markets PI/open or Prime INFORMATION, but they both still exist and can both stil...

... or the DEFFUN declaration for it has been omitted from your code.

This is a "circular" log, retaining only the MAXERRLOGENT most recent entries.

DSGetLinkInfo() or DSGetStageInfo()