Can column layouts be defined dynamically?

dineshrk · Post by **dineshrk** » Wed Feb 04, 2009 9:25 am

As per my requirement, both the input and output file layouts are dynamic
for example if 100 columns are there in the input and 50 columns are wanted in the output today,

Tomorrow there may be 120 columns in the input and 70 columns in the output.

Can we incorporate this in datastage?

**Note - subject changed to be more descriptive - Content Editor **

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Wed Feb 04, 2009 9:31 am

Can do.

How do you intend to find which columns are required in output?

ray.wurlod · Post by **ray.wurlod** » Wed Feb 04, 2009 2:58 pm

Get "them" to justify the requirement. As Sainath asked, WHICH columns will be required in the output on any particular run, and how are you expected to know that and to convey that information to the DataStage job? The RMM stage* has not yet been perfected.

* Read My Mind

dineshrk · Post by **dineshrk** » Thu Feb 05, 2009 6:55 am

Sainath.Srinivasan wrote:Can do.

How do you intend to find which columns are required in output?

they will place the input file and the output file layout (i.e) the column names required that day in the same path

betterthanever · Post by **betterthanever** » Thu Feb 05, 2009 7:21 am

you can mention schema of the input/output file dynamically in a sequentail file stage

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Thu Feb 05, 2009 7:24 am

If there is no transformation, you can use schema files. But your requirement appears quite vague.

Why use datastage to just pick a few columns? I cannot see any step before or after this activity.

dineshrk · Post by **dineshrk** » Thu Feb 05, 2009 8:07 am

Sainath.Srinivasan wrote:If there is no transformation, you can use schema files. But your requirement appears quite vague.

Why use datastage to just pick a few columns? I cannot see any step before or after this activity.

Actually there is no transformation. It is almost like filtering out few different columns everyday.

Sainath.Srinivasan · Post by **Sainath.Srinivasan** » Thu Feb 05, 2009 8:09 am

Then why not control in the source itself ? Is it out of reach?

What is achieved by this extraction? i.e. what happens after this extraction?

betterthanever · Post by **betterthanever** » Thu Feb 05, 2009 8:11 am

Actually there is no transformation. It is almost like filtering out few different columns everyday.[/quote]

then you should be good mentioning the layout as schema file in both input/output..