M
Marco
Hello there!
I'm running IIS 5.0 on a Win2K Advanced Server system which will offer
access to some advanced calculators through a web service interface
(there's no web site on the system, just web services).
In order to profile the connection time I've written a simple program
which opens a socket with the server, sends an HTTP message containing
a SOAP request (for a method which multiplies its two double
arguments) and fetches the result.
Interestingly enough I've noticed that the server seems to be
responding to the first call much faster than to the following ones.
Example output:
<pre>
Result = 12.000000 - Received in 0.022101 secs.
Result = 12.000000 - Received in 0.203185 secs.
Result = 12.000000 - Received in 0.201810 secs.
Result = 12.000000 - Received in 0.199868 secs.
Result = 12.000000 - Received in 0.205943 secs.
Result = 12.000000 - Received in 0.210548 secs.
Result = 12.000000 - Received in 0.209605 secs.
Result = 12.000000 - Received in 0.209832 secs.
Result = 12.000000 - Received in 0.210073 secs.
Result = 12.000000 - Received in 0.201541 secs.
Avg. response time: 0.187451 out of 10 sims.
</pre>
(the client is running on a Unix system; time is measured by
gettimeofday)
The behaviour persists no matter whether connections are kept alive or
closed everytime and no matter whether sockets are preserved ore
uniquely assigned to following requests.
Any suggestions? Any useful reading concerning web sevice performance
optimization?
Cheers
-- Marco --
I'm running IIS 5.0 on a Win2K Advanced Server system which will offer
access to some advanced calculators through a web service interface
(there's no web site on the system, just web services).
In order to profile the connection time I've written a simple program
which opens a socket with the server, sends an HTTP message containing
a SOAP request (for a method which multiplies its two double
arguments) and fetches the result.
Interestingly enough I've noticed that the server seems to be
responding to the first call much faster than to the following ones.
Example output:
<pre>
Result = 12.000000 - Received in 0.022101 secs.
Result = 12.000000 - Received in 0.203185 secs.
Result = 12.000000 - Received in 0.201810 secs.
Result = 12.000000 - Received in 0.199868 secs.
Result = 12.000000 - Received in 0.205943 secs.
Result = 12.000000 - Received in 0.210548 secs.
Result = 12.000000 - Received in 0.209605 secs.
Result = 12.000000 - Received in 0.209832 secs.
Result = 12.000000 - Received in 0.210073 secs.
Result = 12.000000 - Received in 0.201541 secs.
Avg. response time: 0.187451 out of 10 sims.
</pre>
(the client is running on a Unix system; time is measured by
gettimeofday)
The behaviour persists no matter whether connections are kept alive or
closed everytime and no matter whether sockets are preserved ore
uniquely assigned to following requests.
Any suggestions? Any useful reading concerning web sevice performance
optimization?
Cheers
-- Marco --