DSXchange

On the Conditions page, where you set the rules for Conditional Lookup and Lookup Failed, there is also a dropdown list of reference input link names. You can choose to allow multiple rows to be returned from one reference input link. If you want multiple rows to be returned from two reference input...

The SCD stage takes care of all of that, by using memory-mapped copies of the dimension tables. Read Vincent's blog or the IBM collateral to see how it's done.

IBM has (had?) a class (code DX506) called RTI/Web Services.

Might be worth trying to track that down.

Yes, but the Num() function is easier.

Look at the job score and find where the buffer operator was inserted. My guess would be on the inputs to the Funnel stage. Start your diagnosis based on what you find in the score.

I ignored that because the post was made in the server forum.

A field at a time.

Do you have inter-process row buffering enabled? If you do, then four transfomers will be better than one, because they can take four process rather than one. Without inter-process row buffering enabled four directly-connected transformers will execute in the same process, and probably burden that p...

Remove duplicates from the reference input, or configure the Lookup stage to permit duplicates.

How do you execute Teradata macros in SQL (or in BTEQ)? You use exactly the same techniques - embedding them in SQL statements typically - if using them from DataStage.

So you're not using an ODBC database stage, you're using an ODBC Connector. Is this correct?

Compare the column definitions stored in the Connector with those used in your job. It is these differences that the connector can not reconcile.

Get rid of the Oracle stage.

Links that are optimized out of the job design will never show row counts. For example, the output of a Copy stage that makes one identical copy of its input will never report row counts, because the Copy stage is optimized out when the score is composed. Therefore the Copy stage does not exist to r...

Are the hashed files identically configured (see Options in Hashed File stage)? Is the 64BIT_FILES setting in uvconfig the same in both machines?

Add a suitable WHERE clause. Get the most recent update time from the target and use that as a job parameter, which you add to the WHERE clause.