do you usually keep partitioning on "Auto" in your

[dsex100]

Hello,

I have experience with the Server version of Datastage.
I'm trying to get the partitioning flow of Datastage enterprise. "px"

I have an interview.


I notice that the Partitioning is set to "Auto" for the stages and to me this does imply that a developer can just let Datastage do the work.


So, I would like to know if this is in fact how YOU work with Datastage. Do you just leave it on "Auto" or do you in fact have to manually change this for the project you're working on?

The main reason that I can see to change this would be for performance reasons. Like if the data set would be greater than the RAM.

(So if you did change the partitioning from auto to your own manual setting how did you determine if the this was necessary?

I mean did you just see that the job was too slow in Director and then go back in redo it, or did you really manually calculate based on data set volume and amount of server RAM?)

I would be very interested to hear your experience.

[mikegohl]

I would recommend that you read the Parallel Developer's Guide.

[ray.wurlod]

DSXchange Learning Center has a server to parallel transition class available either in DVD format or as an instructor-led class.

Both IBM's and my advanced classes address the issues of best choice of partitioning algorithm.

[John Smith]

If you have an interview then best not to BS if you're out of your depth.
Partitioning is pretty fundamental in PX jobs so you will be caight out if the interviewer knows his stuff.
If there is time prior to the interview to take the DVD course as Ray suggested that is your best bet.

[dsex100]

thank you,..

I actually found what I was looking for in the Parallel Job Advanced Developer's Guide.

[dsex100]

thank you again, and I am expecting the Server to Parallel DVD course to arrive shortly..

I am definitely looking forward to it.

Also thanks Ray Wurlord for informing me that will be part of the course.