Taking DataStage jobs and running them in MPP environment

[AhmadAliFilsoof]

Can you take a Server job and run it in an MPP environment?

[kcbland]

No

[AhmadAliFilsoof]

Ken, thank you.

Can you take a server job and run it in parallel extender in an MPP environment then?

thanks again.

[kcbland]

My original answer was wrong, so I have edited it out. I'm now back to I'm not sure. However, no matter what, you're facing having to change your job to fit the model required to run a Server job as a container inside the parallel framework. You will lose a lot of the functionality that make Server jobs so flexible.


The easier path is usually to rewrite your job as a Parallel job and be done with it.

[ray.wurlod]

The answer is a qualified yes. You have to encase the server job in a shared container, which you can then use in a PX job design. Then the resource file governs its use in an SMP or MPP environment.
The qualification is that there are some restrictions on this (for example that the server job does not use COMMON). It's well described in the Parallel Job Developer's Guide, page 2-17 (Incorporating Server Job Functionality).

[stewarth]

Actually the answer is NO!! DS Server jobs or Server shared containers can only be run on a SMP platform regardless of if PX is used or if a shared container is used within a PX job. A pure PX job can be run on SMP, Cluster and MPP hardware with little to no effort.

Stewart

[ray.wurlod]

The answer changes to yes at DS 7.0!
Although the manual specifies no for MPP at release 6.x, at least one person at Ascential claims to have it working, in another thread in this forum.