DSXchange

I have a sequential file stage which goes to a transformer. One link out of the transformer goes to an Oracle stage and another goes to a pivot stage where I pivot 60 columns 9 times. I am getting very poor performance on this. If I remove the pivot stage and run the job it runs much faster. Anyway ...

Hi there,

What is an optimal setting for this?
If the answer is, it depends then what is the process for finding the optimal setting?

Do I continually increment it until I see no performance gain?

Thanks for the help!!

Yes. I changed the metadata in the columns page to reflect the change. However when I run the job, it does not recognize the change since the change that was make is actually a key in the change capture stage.

Yes, my column's metadata changed from varchar2(2) to varchar2(4). What about if I am trying to maintain a type 2 SCD? I would want to expire the current records and insert the new ones. With the change capture stage this would not happen because the column that changed is part of the change key. Th...

I have a table where the key value has increased from 2 to 4. When I run the job it does not update this change and you cannot add it to the values property in the change capture stage, it errors and say it cannot be both a key and a value. How do people here handle this kind of change?

TIA.

Sean

Yes.

Same problem I have.
Still do. Engineering provided me a patch, but when I install it, it does not work. I have not heard anything new from them in quite some time now.

I would report it.

Sean

Implicit should work. I have a file that has a variable piece to it. In front of the variable pieces are the legth of that field. By specifying that byte as the prefix, DataStage will know how to handle it.

Try the .o library instead and define that in DataStage as an object.
Let me know if this works for you.

I could not get a .so to work on Linux but the .o does!

Sean

When compiling make sure to compile as a release not debug or else it will not work. I never got a dll to work. I compilied the code as a .lib and this works fine for me.

Let me know if you need more info.

How about a where clause on the partition key column. This will be very efficient.

I have a partitioned table that is partitioned on column extract_date. When I select out of it I select where extract = YYYYMMDD.

Hope this helps.

Thanks for the reply,

Is there a way for DataStage to do this kind of operation?
Or is utilizing the database the best way?

I have 2 datasets in an Oracle database that I want to combine. So for example let's say table 1 has the following values:

A
B
C

and table 2 has the following values:

D
E
F


I want the result dataset to be

A
B
C
D
E
F

What is the best way of going about this?

Thanks

I traced this down to a patch IBM support provided me to fix an issue with tranformers with a large amount of columns. Once I restored to a pre-patch state, no problems.

FYI

This one is strange... I am trying to do a calcualtion in a PX transformer. In the transformer I am applying the formula... (x/y) * z where x is a decimal 13,2 y is an integer and z is an integer The result variable is defined as 13,2 The numbers are x = 00000167037.66 y = 181 z = 181 The result sho...

I am investigating ways to create libraries that can be re-used. I was wondering if there is anyway to make a build-op dynamic. That is can we pass it any columns in and process them? It looks like a custom stage is dynamic enough since it has a mapping tab. As long as you have a orchestrate operato...