[Zope-dev] Medusa Request Queue/profiling

[Brian R Brinegar]

Good day,

We are in the process of profiling our web cluster which runs ZEO. One
thing I am interested in is what happens when a Zeo Client recieves more
requests than it has threads to handle. Our Zeo Clients have 8
threads running. It is my understanding that if more than 8 requests are
sent to a client at one time it will queue up the remaining requests
and process them as threads become available. Is this correct?

If so is there a way to easily see how many requests are queued on each of
our Zeo Clients?

We would like to increase the overall speed of our system. And my first
intuition is to increase the number of threads. I think on average
we have more requests than we have threads to handle but I would
like to see how often requests are queued on the Zeo Clients to
confirm this. Ideally we would have enough threads to handle our average
load of requests (I think). However if adding more threads to our existing
machines is going to require more memory than the machines have it's time
to add/upgrade machines.

If anyone has comments, suggestions, or knows how to find out some of the
numbers I'm looking for please let me know.

Thank you,
-Brian Brinegar