With JMeter and the HTTP Samplers (normal and HTTPClient), the latency is the time (in ms) for receive the first 8192 bytes. If response is less 8192 bytes, latency is the time for entire response.
Pure processing time must be getting directly from Webservices, for example with a log (example: in Java/J2EE with log4j with NDC behavior). You can get a transmission time (seconds or microseconds) from Apache Httpd, Tomcat/Jboss, WebSphere, etc. with custom pattern of access logs. This transmission time include the processing time (for backend) + network time for response without last packet (tcp). Milamber Le 23/11/2010 10:52, sebb a ecrit : > On 23 November 2010 10:33, Albrecht Weiser <[email protected]> wrote: > >> Hi Jmeter users, >> i have created a Jmeter test scenario. I'm testing several Webservices >> there. I want to get the PURE PROCESSING TIME of the services itself >> (without network losses, etc.). Jmeter offers two times to get with the >> listeners: >> 1. Time needed (I think it's overall time) - t >> 2. Latency time (?) - lt >> If googleing, the statements to latency time are very general. From my point >> of view, i need the latency time, if the processing time of a service is >> ment. >> Can anyone verify my assumption? >> > http://jakarta.apache.org/jmeter/usermanual/glossary.html#Latency > > >> Bye >> Albrecht >> >> -- >> Albrecht Weiser >> mailto:aweiser AT gmx.de >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected] >> >> >> > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
