parallel query dequeue wait

This shows up as the absolute king of the wait list when I do large imports  where tables have a degree > 1.

As the imp process finishes loading data and begins building indexes, there seems to be a very long period of time where the parallel query servers just sit and wait, not doing any I/O. However, during this period, there is quite a lot of I/O ( writes ) that isn't associated with any of the non-background processes. These writes are not redo or rollback writes.

What is going on here? Does the coordinator need this much time and I/O to calculate who does what?

Anything I can do to speed this process up?

Thanks much,

-Kevin Received on Sat Jun 20 1998 - 09:25:30 CDT