PX Routines

[thumsup9]

Hi,

I have a situation where a job design has data transferred into a staging area.We use a shared container where we have the logic for the rowcount on the records.This Shared container is used inmany other jobs.
Now I would like to know,if a Buildop or Px routine is better than a shared container,which to me seems right.If yes,can you suggest me how we can approach this.Or is there any other way around which isnt striking me,as its only the rowcount logic involved.

Thanks,
Thums

[vmcburney]

I think it is better to do row counting after the job has finished. Trying to embed row counting into the job only slows it down and adds a development and maintenance overhead. There are plenty of ways to do row counts, see the FAQ forum thread on DataStage reporting options such as MetaStage and ETLStats.

[ray.wurlod]

I agree 100% with Vincent on this one. Remember the adage about performance; measurement should have minimal impact on the object being measured.
DataStage preserves the link row counts, so there's no good reason to be interrupting its real processing to enquire while the job is running. Do it afterwards!

[jasper]

I use an after routine for this. most important statement in this:

Code: Select all

 CountRows = DSGetLinkInfo(DSJ.ME, Stage, Link, DSJ.LINKROWCOUNT)

.

This gives you the number rows that were processed on this one link in a stage.

[thumsup9]

Thanks All, I will test based on your suggestions and get back to you with the results.