Preserving the Sorting after FILTER stage

[parag.s.27]

We are using the Oracle10 g as our source database. When we extract the data using OCI stage we sort the data by applying "Order By asc" on Keys and the Date at the end of query. After the data is extracted, we split the incoming data into 2 streams on the basis of dates where if records are having Date < 01-01-2005 00:00:00 then it should go to "Stream 1" where as if records are having Date >= 01-01-2005 00:00:00 then it should go in "Stream 2".

Now in Stream 1 we apply Remove Duplicate stage on the Keys and retain the last record.

Is it possible that after spliting the data the Stream 1 may get records in Un - Sorted manner and thus the Remove Duplicate stage will yield incorrect results?

[nani0907]

Use sequential mode Execution which makes sure that data flows on one node

[dhanashreepanse]

It is always better to sort the data before giving it to the Remove Duplicates Stage.

[mahadev.v]

Sort order might remain but i would have Removed the sort from the SQL and done it in DataStage on stream 1 before the remove duplicates stage.

[parag.s.27]

Actually we are getting review comments from the external contractors who are working as a DataStage professionals about removing the Sort stage. They are saying that Sort stage is not needed as the partition will be preserved.

[ray.wurlod]

Is it possible that after spliting the data the Stream 1 may get records in Un - Sorted manner and thus the Remove Duplicate stage will yield incorrect results?

No.
There is nothing in what you are doing that would affect the sorted order.

[ray.wurlod]

Actually we are getting review comments from the external contractors who are working as a DataStage professionals about removing the Sort stage. They are saying that Sort stage is not needed as the partition will be preserved.

Partitioning and sorting are not the same thing. The comments are correct, but miss the point that a Remove Duplicates stage expects sorted input for most efficient operation.