It doesn't answer your question in general, but I increased mine from
300 to 900 threads on a dual-processor, dual-core machine, and saw
significantly increased throughput with short, simple requests. I
didn't try any in-between settings, though, so I don't know if there
might be a sweet spot somewhere between those settings, and I don't know
if going higher would help or not, because the setting it's at right now
handles the full load with no waiting.
D
[EMAIL PROTECTED] wrote:
Hi,
We have an architecture where we have many tomcat app servers load
balanced by apache at the front.
In resolving our current bottleneck i used JProfiler to see what the
tomcat applications were doing and when under high load there are a lot of
threads which are blocked on this:
org.apache.tomcat.util.threads.ThreadPool$ControlRunnable.run()
This suggests that they are spending all their time waiting for tomcat to
run them, either way they're not runnable anyway.
Our app pushes through huge quantites of very quick requests - typically
each request takes a total of 50ms. We have a limit of 300 connections on
our AJP connector.
The point of blocked threads occurs way before we hit that limit. At the
same time neither the machine is busy, nor are there any obvious problems
with memory usage for the instance.
And indeed; Adding more app servers keeps the load to each one down and
does resolve the problem - i.e. we can increase throughput.
So; My question is, is there a reccomended limit to how many threads
tomcat can handle before it spends more time managing threads than it does
doing real work?
Thanks!
Dan
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
