I have a few more questions/comments about how JMeter measures response times.
Suppose I have the following Test Plan structure that is set to run twice by one user:
Test Plan
Thread Group
HTTP Cookie Manager
Simple Controller 1
HTTP Request A (Retrieve all embedded resources)
HTTP Request B (Retrieve all embedded resources)
Timer - 5 seconds
Aggregate Report 1
Simple Controller 2
HTTP Request C (Retrieve all embedded resources)
HTTP Request D (Retrieve all embedded resources)
HTTP Request B (Retrieve all embedded resources)
Aggregate Report 2
1. There doesn't seem to be an option for JMeter to not retrieve embedded resources
once it has been retrieved the first time. In other words for JMeter to simulate a
browser cache. In my example, the first loop will retrieve all embedded resources for
all the requests. The second loop through would do the same for the same user.
Furthermore, it would be nice during the first loop, when encountering Request B again
in Simple Controller 2, that it would not retrieve embedded resources for Request B.
So it would seem that the response time for the requests will always be measuring
"first cache hit" scenario.
2. Peter, you said that JMeter measures the total elapsed time. Just to
clarify....For example, on a HTTP Request A that is run only once, does the
measurement time start when the first byte request is sent and the measurement time
stops when the very last byte response is returned which could well be the byte from
an embedded file? Or is it the measurement starts after a virtual user has sent the
last byte of the request and ends when the user retrieves the first byte of the
response.
3. The Timer under the first HTTP Request B does not get calculated to the response
time for Aggregate Report 1, which is what I want. However, if using the Transaction
Controller, it does. Transaction Controller should have an option to measure with
Timers or not.
4. I suppose JMeter currently does not measure the number of KB a page
request/response takes? Is this planned to be implemented?
Thanks,
Mabel
Re: Jmeter accuracy (applets, load, response)
--------------------------------------------------------------------------------
From: peter lin
Subject: Re: Jmeter accuracy (applets, load, response)
Date: Wed, 12 May 2004 10:58:11 -0700
--------------------------------------------------------------------------------
most of this is in the component documentation.
http://jakarta.apache.org/jmeter/usermanual/index.html
the response time you see in JMeter is the total elapsed time. If you do not check
"retrieve all embedded resources from HTML files", it is just the html. If you check
it, it is the total time to get all the resources including html.
Remember the response times are always estimates and never use the average as "actual
response time" from a user's perspective. Since JMeter does not render the pages, the
response time is faster than the total time to transfer the data + browser rendering
time. Depending on how deeply nested the html tables are, it could take the browser
and additional 2-5 seconds to draw the page.
keep in mind network latency can greatly affect the user's perception of response
time. If someone is using a bad modem, the page will load considerly slower. I tend to
error on the side of safety and state the results are only indicators of likely
performance :)
you'd be amazed at how people will interpret the results.
peter
Steve Luong <[EMAIL PROTECTED]> wrote:
Hey guys,
I was wondering how accurate is Jmeter in measuring load time. I'm working
on a report for a client and I need to know are these numbers reliable.
I've only been working on Jmeter for a few days and everything looks
promising. A few questions I would like to know is are there any hints
anyone can provide to give me the most accurate information on the load
times. Also when Jmeter is set to follow all elements in the html page like
images does it do applets too? Does it measure the complete time it takes
to download the applet or just the time it takes to send the applet? Having
this knowledge would greatly help in my efforts. Thanks for all the help!
SL
---------------------------------
Do you Yahoo!?
SBC Yahoo! - Internet access at a great low price.
