transformer stage

[cetzhbo]

Hello Gurus,

for transformer stage, there is one input link, and serverial outputlinks, why is it not designed to include more input links ?

thanks very much!

[ArndW]

It can have only 1 main input, and n-output links. If it had more than 1 main input, then it would have to be merge or join stage.

[ArndW]

It can have only 1 main input, and n-output links. If it had more than 1 main input, then it would have to be merge or join stage.

[cetzhbo]

thanks. there should be some reason to use transform function by transformer stage/merge stage/join stage. please tell me why?

[cetzhbo]

I mean why don't comine the function for transformer/merge/join into one stage, an we can configure this stage in different way to get different function ?

thanks

[ArndW]

In DataStage server jobs the transform stage additional can do lookups. In PX each stage performs a specific task and you end up using more stages, but each one is specialized and tuned to it's specific task.

The Transform stage cannot be configured to do more (or less), that is just the way the application works.

[cetzhbo]

thanks, I got it

[samunik]

I mean why don't comine the function for transformer/merge/join into one stage, an we can configure this stage in different way to get different function ?

thanks

This is a classic learning curve from Server to PX (been there). But in order to understand why it is this way, lets revisit the history. Although both PX and server are under the same umbrella of IBM and before that Ascential but these 2 products have very different origins Ardent made Server and PX engine came from Torrent. Server is written in BASIC and PX in orchestrate. Hence in PX all the batch related functinalities that can be performed were broken out in stages to better optimize the services. Ascential just enhanced the functionality and so did IBM.

Gurus is this correct or a bunch of jibrish.