DSXchange

deepthi wrote:What is the maxium value that this key_mgt transform generates upto?
Will it resets after the number is reached?

Thanks
Deepthi

No practical limit, and it does not reset. The engineering limit depends on the precision setting, but you can always get at least a fifteen digit number.

You will have to request it from your support provider - DSXchange can not provide, or (legally) can any of its participants.

With a text file you have the option of sqlldr.

But with small files, and presumably given that all you're going to do with the file contents is to call UtilityRunJob(), I think a Folder stage may be worth considering.

Remember that a dependent pass uses only the residuals from the previous pass as source.

Which particular Oracle stage are you using?

You should find Open Command and Close Command properties under the Available Properties to Add frame when the Options folder is selected, if you are using the Oracle Enterprise stage.

It really is a different question warranting a separate thread. Is there a message logged showing what the UV stage has done? What is the data type of the Key column?

Load the tables into local text files. Use cat to put them together for UNION ALL. If you need to remove duplicates, there are several strategies available, all of which need the file to be sorted. And, of course, tune your database queries. Make sure that the tables have been analyzed so that the q...

Welcome aboard.

You might like to investigate the Folder stage.

Use a sparse lookup and specify the SQL in the database stage.

And your key performance indicators are... ?

Why bother? Why not create your FA field with the starting position of the first field you need and the ending position of the last field you need?

An inner join (those rows that exist in both sets) can never be the same as a Cartesian product (all rows from the right input joined to every row from the left input). If the left input contains N rows and the right input contains M rows, then an inner join will return the smaller of N and M rows, ...

Maybe. But the locks aren't held for long, so maybe allowing each job to WAIT for any lock on the sequencer to be released would be a sufficient solution.

Start by defining "performance" in an ETL context.

Hint: rows/sec is definitely NOT a meaningful measure of anything.