DB2 Performance Question

Havoc · Post by **Havoc** » Tue Sep 11, 2007 7:28 am

We are reading from a DB2 Table which has not been partitioned. This table is being read by a DB2 UDB API Stage (with a node constraint set to node1) . Will there be a performance improvement if we use a DB2 UDB Enterprise stage in reading this table ?

I have read the article where it mentions there is a 250% performance improvement over the API Stage but noticed that it was done on a partitioned table.

Can someone please provide their insights on this ?

Thanks in advance

ray.wurlod · Post by **ray.wurlod** » Tue Sep 11, 2007 4:13 pm

No improvement if you continue to use SQL. The DB2 Enterprise will see the table as having a single partition, and therefore operate in sequential mode.

Havoc · Post by **Havoc** » Wed Sep 12, 2007 9:59 am

Okay, I just tried this ....
the read speed from the Enterprise stage tripled! .. Is there an explanation to this ?

ray.wurlod · Post by **ray.wurlod** » Wed Sep 12, 2007 8:21 pm

What else was happening during the API test?
Was the DB2 Enterprise stage set up to use direct read or SQL?

Havoc · Post by **Havoc** » Wed Sep 12, 2007 10:14 pm

ray.wurlod wrote:What else was happening during the API test?
Was the DB2 Enterprise stage set up to use direct read or SQL? ...

Nothing else was happening during the API test. Enterprise stage was set up to read the same SQL. The same SQL as in API stage which involved a join statement.