DSXchange

Hi,
I am working on 7.5.1 EE. I am designing parallel routines. We have so many partitioning methods like hash, auto, DB2, round robin etc. Can anybody tell me in what circumstances which partition we have to use? Are there any specific rule to use a particular partition?
And also can anybody tell me what performance issues we have to take care of and how to resolve them?

If you can provide any links for it. It will be a great help for me.

Thanks
Kavuri

If you are joining two (or more) data sources using a Join or Merge stage, sort and then hash-partition all the data sources on the join key. Doing this will ensure that all the records from all the data sources that share common join key values will end up in the same partition.

For performance reasons, avoid re-partitioning your data if you can.

[quote="kavuri"]Hi,
I am working on 7.5.1 EE. I am designing parallel routines. We have so many partitioning methods like hash, auto, DB2, round robin etc. Can anybody tell me in what circumstances which partition we have to use? Are there any specific rule to use a particular partition?
And also can anybody tell me what performance issues we have to take care of and how to resolve them?

If you can provide any links for it. It will be a great help for me.

Thanks
Kavuri[/quote]

Thanks Jamach,
i.e we need to opt for preserve partitioning and use same right?

Any other comments are welcome and any more performance issues please let me know.

Thanks
Kavuri