Re: 2-node 9iRAC --> 2nd instance performance problem

This makes absolutely no sense.
All of them are idle messages.
You are measuring the wrong wait events.

--Steve

ernestsiu_at_yahoo.com (Ernest Siu) wrote in message news:<2833144d.0403170917.337e1053_at_posting.google.com>...
> What does it mean by the increased message wait time given that they
> are the same query?
>
> Ernest
>
> ernestsiu_at_yahoo.com (Ernest Siu) wrote in message news:<2833144d.0403151631.13528fc9_at_posting.google.com>...
> > One key thing I left off is: I'm using a WAN emulator (BSD DummyNet)
> > between the two 9iRAC nodes, however with only 0ms delay (yes, I
> > intend to run some distance test later, but I don't expect it to have
> > trouble with 0ms). No distance element at storage level. I know this
> > is out of the ordinary, but I believe it should at least work with a
> > 0ms emulator (given that the emulator works properly). I suspect
> > there's something wrong with the DummyNet box in terms of network
> > setting....
> >
> > I'm looking into 3 items, the time_waited for: gcs remote message, ges
> > remote message and rdbms ipc message. Performing the same query of
> > SELECT 1000000 rows:
> >
> > ges re-msg gcs re-msg rdbms ipc msg Time-to-perform
> > 0ms Bad Box 848132 1691930 5918365 8480 seconds
> > 0ms Good Box 5808 11455 35281 56 seconds
> >
> > The time-waited is proportional to the actual performance. Hmm... I
> > suspect there's something wrong with the network setting on the
> > DummyNet that it doesn't like. What could it be: MTU, Window-size,
> > Duplex? Or the setting on the 2 hosts?
> >
> > Ernest
Received on Mon Mar 22 2004 - 02:32:03 CST