DSXchange

Hi all,

I have rewritten a server job with a parallel job for better performance aspects. This job basically reads a sequential file and performs 5 different lookups and write the resultant data into another sequential file.

I am using Oracle connector stages for lookups as against DRS stages in Server job. We are running on 4 nodes. Currently for 'Enable partitioned reads' property, the value is No. Just wanted to see what the difference would be in terms of performance if I make 'Enable Parttitioned Reads' as Yes. I see that there are diffrent partitioned read methds as well - Rowid range, Rowid hash..etc. What should be considering if I untend to make the partitioned read as 'Yes'. Please let me know your inputs. Thanks!

I've only used the feature with the DB2 Connector, but you want as even a distribution across the processing nodes as possible. Experiment with the different options.

Mike