Niall >showing me that if you change a parameter the plan will change tells >me very little about performance - it does tell me about a bit about >behaviour, showing me response time does and - so far as I can tell - >there aren't any examples of improved response time from setting these >parameters. Its as if the tuning goal has become 'favour nested loops' >rather than 'impove end-user experience'. In the example I posted some hours ago shows that: SQL> ALTER SESSION SET OPTIMIZER_INDEX_CACHING=3D0; SQL> SELECT count(*) 2 FROM t t1, t t2 3 WHERE t1.id =3D3D t2.col1(+) AND t2.col1(+) > 900; Elapsed: 00:00:00.07 SQL> ALTER SESSION SET OPTIMIZER_INDEX_CACHING=3D90; SQL> SELECT count(*) 2 FROM t t1, t t2 3 WHERE t1.id =3D3D t2.col1(+) AND t2.col1(+) > 900; Elapsed: 00:00:01.05 This means that just by setting OIC there is a difference of factor = *15*. Chris -- //www.freelists.org/webpage/oracle-l