i have a question about performance tools.. while these tools execute performance test, they using proxy setting.
what is the logic of this situation? why do they using proxy?
what will be happen if they don't use?
and, what is the structure of that?
If you do this, then you need to document the heck out of this inclusion and fully disclose it along with your results. Who is to say that the proxy server won't become the bottleneck in your test and limit the scalability of your test solution?
It has happened before.
If this proxy scalability issue does happen in your case it will be to your benefit to catch the issue before a developer finds it when his/her code is being challenged by an apparent lack of scalability as measured by your test
These are excellent points being made about the use of proxy servers in simulations. However, I think that mahsum may possibly be referring to proxying being done on a load client.
The load tools (and other http examination tools) I have worked with use a local proxying tool on the recording machine to capture web traffic. This seems like a suitable method for recording/other singleton exercises that do capture. The logic is that this provides a handy way to sniff requests without the browser components having to be involved at all. If these proxy settings weren't in place, the program being used to capture http wouldn't work, because these programs are designed to capture requests as they pass through the local proxy.
Note: This can be an issue if the proxying program crashes; the proxy settings are not restored to previous, so I have had to go into browser configuration before to restore these.
My primary load tool (QALoad) does not change the proxy settings during playback, but I could not attest to what other tools do. I could see how this might be a playback strategy. Are you sure these settings are changed during playback?
Or, to bring it back around to A$h and James, are you certain that these aren't the proxy settings for your test environment?