Performance problems with distributed queries with table joins.

I am wondering if anyone has found performance problems with distributed queries with table joins. The local Oracle database is running server 7.0.16 and SQL*Net is version 1 on a SUN. The remote database is version 7.0.16 and is also running SQL*NET version 1, but is on an HP.  

The query I am running from the local database is something like

select col1, col2, .....
from TABLEA_at_db_link, TABLEB_at_db_link
where TABLEA.col1 = TABLEB.col1
and TABLEA.col1 = 1;

where tables A and B are both on the same remote database in the same SCHEMA. The field col1 for the joins is a primary key for both tables. The query only returns one row.

On the remote database I ran sql_trace and it appears that ORACLE is doing a full table scan for the first table instead of using the index. What is interesting when I run the same query from another ORACLE database but on an HP server the query take only a couple of seconds to return.

There is nothing special about the query except that it executed is over a link and I do not believe the problem is due to network traffic but to the way ORACLE is executing the query on the remote database.

I would appeciate information concerning this problem  

Thanks
Wayne Kirkcaldy
Edmonton, Alberta
Canada Received on Tue Jan 30 1996 - 00:00:00 CET