Simulating real browser behavior

["Christensen, Alan" <al...@hp.com>]

I have used Jmeter in the past to simply download html pages.  I have
never dealt with the problem of downloading embedded content.  I know
that the http sampler will allow me to force the download of embedded
content by checking a box at the bottom.  However, this does not result
in behavior that is similar to the way that browsers work.  In
particular, Jmeter appears to request these items one at a time in
serial fashion.  This is not the way that browsers work.  Browsers often
request many content items in parallel.  The number of parallel
retrievals is dependent upon the browser settings and where the content
items are located.  Is there a way for Jmeter to work similarly, or does
its structure require it to request items one at a time?  If there is a
mechanism to do this, does it preserve the ability to measure the
performance of the entire transaction, i.e., from the first request to
the final piece of embedded content arriving back at Jmeter?

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org

Re: Simulating real browser behavior

[sebb <se...@gmail.com>]

On 06/12/06, Christensen, Alan <al...@hp.com> wrote:
> A second question along the lines of the one below.  Is there a way to
> easily simulate cached behavior in a browser?  If you check the box at

Not at present. There is an outstanding enhancement request to support caching.

> the bottom of the http sampler, it appears that all embedded objects
> will be retrieved every time the sampler executes.  In a real world
> case, many (but probably not all) of the embedded content objects will
> be cacheable.  What would be ideal, would be to force the sampler to
> fetch the cacheable content on the first pass, but not on successive
> passes.   Is there any way to simulate this other than to identify the
> non-cacheable items and include them explicitly in their own samplers?

Not automatically at present.

> If this technique is used, then how would you combine the response times
> for the content items with the response time for the html page to get an
> overall response time?

This already happens for automatic downloads.

If loaded separately, use a Transaction Controller.


> -----Original Message-----
> From: Christensen, Alan
> Sent: Wednesday, December 06, 2006 2:40 PM
> To: JMeter Users List
> Subject: Simulating real browser behavior
>
> I have used Jmeter in the past to simply download html pages.  I have
> never dealt with the problem of downloading embedded content.  I know
> that the http sampler will allow me to force the download of embedded
> content by checking a box at the bottom.  However, this does not result
> in behavior that is similar to the way that browsers work.  In

JMeter is not a browser ...

> particular, Jmeter appears to request these items one at a time in
> serial fashion.  This is not the way that browsers work.  Browsers often
> request many content items in parallel.  The number of parallel
> retrievals is dependent upon the browser settings and where the content
> items are located.  Is there a way for Jmeter to work similarly, or does

No

> its structure require it to request items one at a time?  If there is a

Yes

> mechanism to do this, does it preserve the ability to measure the
> performance of the entire transaction, i.e., from the first request to
> the final piece of embedded content arriving back at Jmeter?

N/A

> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org

RE: Simulating real browser behavior

["Christensen, Alan" <al...@hp.com>]

A second question along the lines of the one below.  Is there a way to
easily simulate cached behavior in a browser?  If you check the box at
the bottom of the http sampler, it appears that all embedded objects
will be retrieved every time the sampler executes.  In a real world
case, many (but probably not all) of the embedded content objects will
be cacheable.  What would be ideal, would be to force the sampler to
fetch the cacheable content on the first pass, but not on successive
passes.   Is there any way to simulate this other than to identify the
non-cacheable items and include them explicitly in their own samplers?
If this technique is used, then how would you combine the response times
for the content items with the response time for the html page to get an
overall response time?  

-----Original Message-----
From: Christensen, Alan 
Sent: Wednesday, December 06, 2006 2:40 PM
To: JMeter Users List
Subject: Simulating real browser behavior

I have used Jmeter in the past to simply download html pages.  I have
never dealt with the problem of downloading embedded content.  I know
that the http sampler will allow me to force the download of embedded
content by checking a box at the bottom.  However, this does not result
in behavior that is similar to the way that browsers work.  In
particular, Jmeter appears to request these items one at a time in
serial fashion.  This is not the way that browsers work.  Browsers often
request many content items in parallel.  The number of parallel
retrievals is dependent upon the browser settings and where the content
items are located.  Is there a way for Jmeter to work similarly, or does
its structure require it to request items one at a time?  If there is a
mechanism to do this, does it preserve the ability to measure the
performance of the entire transaction, i.e., from the first request to
the final piece of embedded content arriving back at Jmeter?

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org