PERFORMANCE ISSUE.

[karry450]

Hi There,


I am new to data stage, I want some solution regarding performance in two cases.

1st Case:

My source is oracle and target is oracle.
Source could be of 4 million rows can anyone please let me know which way is best

I dont have any transformations in between

writing oracle to flatfile and then flatfile
writing oracle to oracle
or using sequential file writing oracle-seqfile-oracle.

or anyother way please.

2nd case

i have two table(oracle) as my source should do lookup on around 10 columns and populate other different columns to target (oracle)


can anyone please let me know which way is best


writing oracle to flatfile and then flatfile
writing oracle to oracle
or using sequential file writing oracle-seqfile-oracle.

or anyother way please.


Thanks
Karthik

[ray.wurlod]

1st Case Option 1
Do the whole move within Oracle. Don't worry about using DataStage.

1st Case Option 2
Oracle ---> SeqFile then use sqlldr with customized ctl file (tuned buffer sizes, etc) to load into Oracle. Obviously this only works if bulk loader is applicable (no bitmap indexes, load mode is relevant, etc.)

2nd Case Option 1
Do the whole move within Oracle.
INSERT INTO target (columns) (SELECT columns FROM table1 INNER JOIN table2 ON conditions)

2nd Case Option 2
Do the join in Oracle then work as for 1st Case Option 2.