DSXchange

A curious suggestion from someone who has "performance driven outcomes" in his signature block!

How many data type conversions (implicit or otherwise) does your "solution" require?

The first code should work if it is the derivation expression for the stage variable svGroupRecordsIdx.

The second code will not work because you have tried to use an assignment statement, but it's actually treated as a Boolean test condition.

"99999999" would take the Else clause of your If statement. You need to add an IsValid() or IsValidDate() test as well.

1) Of these only the first exists. "DB2 connector type" is not an available partitioning algorithm.

2) Try it with an innocuous SQL statement (such as a select of the current date) to determine whether it's executed from each node or only once.

What are you NOW getting as output?

Let's start by asking what you mean by "messed up in Oracle", why you believe this, and how this may have occurred.

Some idea of your DataStage job design might be useful too.

Probably easier to make use of the logs of the sequence and of its dependent jobs. Another possibility would be to take the generated code, paste it into the Job Control screen of a new server job, and add timing points into that code. Examine the code (on the Job Control tab in Job Properties). A s...

The only data types that support precision in your example are the datetime/timestamp and the string/char. Check the precision settings for these in the database tables to which each environment (dev and the other) connects.

Did you unmark the other elements as Key?

You need to mark Amount as your repetition element.

Can you provide the XPath expressions from the Description field?

Actually the test expression deptno = deptno will always be true, so your counter will increment for every row.

Specify Double as the data type the DataStage table definition. Some databases, including Sybase and SQL Server, report Float for DFloat data types.

There are lots of solutions but, without knowing your exact requirement/logic, we cannot comment on whether an out-of-the-box solution exists.

You can call BASIC routines from the BASIC Transformer stage in parallel jobs.

You can use server jobs.

You can rewrite the logic into a parallel routine.

Yes, if you're forcing it to run on two nodes (each of which is performing the SELECT) or, possibly, if you're selecting from a partitioned table. When partitioning by ROWID the constraint can work directly against the row identity value.