Performance suggestion.

pimmit22043 · Post by **pimmit22043** » Tue Sep 02, 2008 8:51 am

Oracle db source has 10,000 rows and joining with 8mill rows table. Even thought it is a parallel job it runs in sequential mode; job is taking more than 2 hrs to complete.
lookup table or source table is not oracle partitioned so we are not partitioning the datastage job and runs sequential mode in parallel job.

Can anybody suggests me how I can improve the performance parallel job.

Thansk for time;
pt

pimmit22043 · Post by **pimmit22043** » Tue Sep 02, 2008 12:26 pm

any suggestions please !

singhald · Post by **singhald** » Tue Sep 02, 2008 12:51 pm

i want to know few more details about the join conditions,

what are the column you are using to join, \are these column have index or not ?. are u performing join in sql or using DS join stage

ray.wurlod · Post by **ray.wurlod** » Tue Sep 02, 2008 3:42 pm

pimmit22043 wrote:any suggestions please !

Learn patience. Your two posts were 11:51pm and 3:24am my time.
How long does the Oracle query take by itself to complete (not to return the first row)?
What is your job design?
Are you doing normal lookup or sparse lookup?
How many rows are there in the reference data? These have to be loaded into a virtual Data Set before the main stream can start.
Would a join in Oracle be more efficacious?