Auto Partitioning

[dsscholar]

Hi,

" Typically InfoSphere DataStage
would use round robin when initially partitioning data, and same for the intermediate stages of a job " from parallel jobs developers guide.

So i went for partitioning logic myself.

Job design:

Source ---> lookup1 (2 reference tables) ----> lookup2 ( 2 reference tables) ---> Target

In lookup1, input link and reference1 are hash partitioned and reference2 is entire partitioned. In lookup2, input link is same partitioned and reference links are entire partitioned. Am getting the following errors.

lkp_EBU_1: Input dataset 1 has a partitioning method other than entire specified; disabling memory sharing.

lkp_EBU_1,5: Could not map table file "D:/IBM/InformationServer/Server/BBPNonHyperion/Datasets/lookuptable.20110728.y51lp5b (size 12288 bytes)": Invalid argument

But if i choose auto partitioning to all the links, the job works fine for me. What partitioning "auto" decides for the above logic. Will auto choose partitioning methods other than "round robin " and "same"?


Please explain auto method for above logic(how the partitioning is done for each link) and why i got those two errors for when the partitioning is not auto.

Thanks in advance

[ray.wurlod]

(Auto) decides Entire for reference inputs to a Lookup stage.