2 remarks about all your stress testing efforts:

A) TIME_WAIT

When not doing HTTP Keep-Alive, under high load the size of the TCP hash
table and the effectiveness of the system to lookp up TCP connections
can limit the throughput you can reach. More precisely, depending on the
excat way of connection shutdown, you get TIME_WAIT states for the
finished connections (without HTTP Keep Alive it could be one such
connection per request). Most systems get slow, once the number of those
connections reaches somthing arounf 30000.

E.g. if you are doing 2000 requests per second without HTTP Keep Alive
and the combination of web server and stress test tool leads to
TIME_WAITs, after 15 seconds your table size might reach a critical size.

The amount of time a system waits before it destroys TIME_WAIT
connections varies. Solaris 4 minutes, but tunable down to 5 seconds,
Linux 1 minute (I think) and still not tunable (a free pizza to the
first one who can show me how to tune the time interval it takes for a
TCP connection to be moved out of TIME_WAIT), on Windows I think also 1
minute but tunable.

Not using HTTP Keep Alive will very likely limit quickly the achievable
throughput when going up in concurrency.

B) Resource usage

You are doing a first interesting analysis, namely the base throughput
you can reach using concurrency one. Now the throughput we can reach
using a given setup always depends on the first bottleneck we hit.

For big files it might be the network bandwidth. For small files it
might be CPU or maybe I/Os per second we can do on the interface.

So it is also interesting to compare the resource usage. Some resources
are harder to measure (like memory), but the one easy to measure
resource is CPU. So you get some more info, when also measuring CPU per
request. Think of it as maximum speed and gas efficiency.

Regards,

Rainer

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org

Reply via email to