My company is having
an OPS configuration, 4xHP 9000/N, HP-UX 11, 64 bit, RDBMS 8.1.7.1, 8
CPUs/node.
I noticed that some
parallel queries are not executed on the node that they were initiated on, but
are migrated
to the
least busy
instance at time. Is anybody else aware of this behavior and is i t documented?
Can it,
somehow, be turned off? A reference to a metablink note would
be very helpful indeed.
Mladen
Gogala
Oxford Health
Plans
Sr. Oracle
DBA