Partition Type = Auto

karthi_gana · Post by **karthi_gana** » Mon Feb 27, 2012 1:38 am

All,

I have designed a simple job as below


Seq_File --> Copy Stage --> Seq_File1
                                    --> Seq_File2
                                    --> Seq_File3

I left the partition type = auto (default value) as it is. I have read somewhere if we choose/left "Auto", the datastage engine will apply either "Same" or "Round Robin". How to know which one will be applied by datastage?

ray.wurlod · Post by **ray.wurlod** » Mon Feb 27, 2012 1:46 am

Search!
This question has been answered in detail before.

karthi_gana · Post by **karthi_gana** » Mon Feb 27, 2012 3:49 am

I did

But I don't find any useful information. May be, i am not using the correct keyword to search it.

karthi_gana · Post by **karthi_gana** » Tue Feb 28, 2012 4:43 am

can you refer some post's to know about it? I was tired to search about this topic. I don't find anything related to my requirement.

Kryt0n · Post by **Kryt0n** » Tue Feb 28, 2012 3:05 pm

For the purpose of your job, what difference does it make which it applies?

For a sequential file, there is no partitioning which is likely to rule out "same"

ray.wurlod · Post by **ray.wurlod** » Wed Feb 29, 2012 12:55 am

Code: Select all

(Auto) ::=  Round Robin if sequential -> parallel
            Entire on reference input to Lookup stage
            DB2 on input link to DB2 stage of any kind (parallel execution)
            Same if parallel -> parallel

karthi_gana · Post by **karthi_gana** » Wed Feb 29, 2012 6:30 am

Thanks a lot Ray.

can you please tell me where i can find more information on this topic?

sequential --> Parallel

do you mean sequential to any parallel stages?

say for example sequential --> filter or copy ?

kwwilliams · Post by **kwwilliams** » Wed Feb 29, 2012 9:23 am

Why don't you choose the one you want instead of using auto? If you're curious as to what it does when you use auto look at the dump score. It will tell you the partition method.

ray.wurlod · Post by **ray.wurlod** » Wed Feb 29, 2012 3:09 pm

Sequential ---> ANY Parallel (except as noted - ref input to Lookup, input to DB2)

More information in the Advanced DataStage class from IBM. Not sure where it exists in the documenatation suite, but it has to be in there somewhere. Have you looked in the Information Center?

qt_ky · Post by **qt_ky** » Wed Feb 29, 2012 6:27 pm

Nearly every section in the Parallel Job Developer's Guide repeats the descriptions of all the partitioning and collecting options. If IBM would pull all that into one section, it may cut the full doc size in half. Read the product manuals, and you can't miss it.