DSXchange

So, you want the first nine rows to appear in every row of output then, from row #10 onwards, the two input columns to appear in column10 and column11 ? Is this a correct specification? What is your target? To get this right you probably need to run in sequential mode. For 50,000 rows that should no...

A node is a logical subdivision of processing resources.

A partition is that subset of the data that are processed on a node.

Ignore rows/sec as indicator of performance. They include the wait time; for most stages the clock starts when the job starts. You can see this clearly using the Performance Analysis tool in Designer.

I have seen the same thing with stage variables. You re-order them and save the job. When you re-open the job, they're back in their original order.
(Version 8.5.0.1)

You probably don't like CurrentTimestamp() either, because it returns the current timestamp (that is, "now") whether or not another record is being processed. I have some DataStage jobs processing over a quarter of a million rows per second on a two-node configuration. "Now" will...

None that you have not already posted. It's the alternate call name. Have a look at any example in DSParams (that is, not the description itself). For example you will learn that function StringToDate has an alternate call name of date_from_string that can be used in a Modify stage.

... and alt_u_call_name is used when Unicode is in effect, and you're dealing with ustring instead of string in the modify operator.

Can you create a trigger on the table that invokes dsjob?

Otherwise investigate the Change Data Delivery product, which can detect the change from the transaction log.

What application causes the row to be inserted? Could that application also invoke dsjob to run your sequence? You may be looking at a multi-instance sequence, with the invocation ID generated from date/time or similar, if inserts are occurring rapidly. Otherwise definitely do investigate Change Dat...

You need a clean mechaism for shutting down the sequence. Therefore you should set a timeout on the WaitForFile activity. If this triggers (Failure trigger = timeout) take a branch to check for the existence of another file that indicates that the sequence should shut down and, if found, invoke a Te...

No, there's only integer to string and string to integer.

You can create new specifications; you can't create new functions.

Back up (copy) the Data Set before starting the job.
Write jobs that don't abort.

Perhaps you should be investigating the Change Data Delivery product.

Does "array DML" in the error message suggest anything to you? What array size are you using?

<oracle ignorance>
Can you use hints in conjunction with array processing? (Try setting Array Size to 1 to answer this question.)
</oracle ignorance>

All three methods use the same tsort operator. By using an explicit Sort stage you get more control over the amount of memory allocated for sorting, and you can generate Key Change columns if that's important to your processing. You also get the ability to handle already-sorted data ("don't sor...