c routine

[jasper12]

Hi,
we now have (in all our datastage-projects) some basic-routines(as used in older versions). Since we now have the latest parallel-version these cause poblems (always have to use a basic-transformer,...). I looked at the c-examples on the datastage CD and what I seem to learn from this is that c-development is a lot harder.
What you need to do:
-create the C-routine
-compile
-create the ds-routine linking to the c-routine
-use this routine in a simple job and only now you will be able to check the code.

Aren't there quicker ways for developping? Or maybe: does someone know where to find some basic-set of routines(which can be found for basic everywhere).
We're talking here about 100 routines that need to be translated, are there other ways of handling this?
What have you done with the old basic-routines, maybe still using them?(for performance it doesn't seem a big difference, we are worried about future releases though)

 

[vmcburney]

You can use the basic routines in a BASIC transformer in a parallel job. If you don't see the BASIC transformer in your list of stage shortcuts then look for it in the repository window by displaying stage types and browsing through the parallel stage types. This lets move the functionality of a Server job transformer straight into a parallel job.

You can also use a server shared container within a parallel job to re-use existing server job logic.

Both methods are slower then a native parallel job as they do not work as effectively across the partitions.

Ascential have a tool that converts server jobs to parallel jobs however they do not make it publicly available. It is possible that it will be in the next release (DataStage Hawk) or available as a service through Ascential.

Your best bet is to start by converting just those server jobs that take a long time, rewrite them as parallel jobs, leave the rest as server jobs. There is no harm in running a combination of server and parallel jobs together, they can be called from the same sequence job, and slowly migrating across to the parallel architecture. If you convert your slowest jobs now you may find the rest are much easier to convert after the next release.

Ascential/IBM will continue to support BASIC transformers and server jobs in the next release and probably for a few years to come.