DSXchange

Define "performance" in an ETL context. Rows/sec is almost totally meaningless (search forum for my reasons). It really depends on how good the ODBC driver is how close it gets to a native driver. As a general rule, however, ODBC needs an extra layer of software (driver manager) as well as has to "t...

=We have the following requirement.

You do not hijack someone else's thread, particularly resolved issues. You create a new topic. Near the top of the main part of the page you will find an icon Post new topic that allows you to do so.

A Data Set has the columns with which it was created. You must delete the Data Set and re-run the job that creates it.

Check permissions for scratch disk direcories, for the UVTEMP directory and for the /tmp directory.

There is an environment variable (something like APT_PM_ROWCOUNTS) that causes each player process to display its row counts.

Welcome aboard.

In parallel jobs you will need multiple TPUMP stages. The GUI tells you this by not allowing multiple inputs.

Server job is a totally different product.

Curious how there's never time to do it right, necessitating allocation of more time to do it again.

Call DSExecute("UNIX", "echo $MYVARIABLE", Output, ExitStatus)

There is a link from Source to Lookup (probably through a Column Generator unless you select the constant key from source). There is no link from Source to Transformer.

Do you have NLS enabled?
Do you have LANG_C or NLS_LANG environment variables set?

Click the Generate Report tool in Designer. Save the intermediate file if desired. Post the report.

The best approach is to change all the job sequences that need changing.

If they won't, use a Distributed file of at least ten parts.

Yes. No. Maybe. Depends on the stage type.

Your question was in the server forum, so got a server job answer.

Do the indices exist? Can you execute DS.REINDEX?