On Friday 18 January 2008 07:17, Johan Stuyts wrote:
> > Odd. Usually people think of excessive CPU utilization as the
> > problem...
> > Does the client submit a request and wait for the response before
> > submitting the next request? If so, the transmission and
> > client-side decoding time produce idle periods on the server.
> ... The client and the server are running on the same
> machine and that is why I expect the CPU to be utilized fully all the
Is it possible that somehow the packets making up the exchange during
the test are traversing a physical, external network instead of going
through an in-RAM short-circuit path?