Page front page load performance test analysis of the feasibility scheme of each tool

Page front page load performance test analysis of the feasibility scheme of each tool

Levy on the various browsers and tools to do the test, the choice is the Baidu home test, because it is relatively single, no multi-layered nesting and continuous loading, minimizing other effects, the test found that some of the current program is not feasible, the following test results give a feasible solution and not feasible recommendations.

1. Comparison of WebBrowser (kernel IE10) and IE10 browsers:

WebBrowser's Test method is to first perform clear browser cache, starting from the start of the navigation, DocumentCompleted time will be 2 times, take the last time, according to DocumentCompleted interpretation, this time equivalent to all elements loa complete , the test results are as follows:

Next Open the IE tool test, first do clear the browser cache test results are as follows

Conclusion: After many tests, we can see that the results of WebBrowser and IE tools test are quite close!

Webbrowser documentcompleted time is the full element load time of IE tool

2 HTTP Grab Kit Fiddler results test:

End time

Clutch IE10 results are as follows:

Start 14:15:59:6241

End 14:16:00:1082 The interval is 481 milliseconds.

Clutch webbrowser results are as follows:

Start 14:18:37:4179

End 14:18:37:8689

They are 451 milliseconds apart.

Conclusion: 1. IE10 and WebBrowser Network load times are basically the same,

2 PROBLEM: Fiddler grab packet data is larger than the browser's calculation time? Is this a fiddler data problem?

2. Firefox's Firebug tool test:

Clear the browser cache first:

And then turn on Baidu

Conclusion: How to test all over 2000 milliseconds, do not understand why more than IE slow so much? Straight eyes can see more slowly than IE!

3. BERSERKJS Test

This doesn't know what the cache is, just test it.

You can see that this interval is 246 seconds, the first screen time is: 1894 milliseconds

This time is similar to IE kernel

Fiddler Test Grab Pack:

14:34:09:7921

14:34:10:2181

426

You can find that the Fiddler grab time is greater than the BERSERKJS test time!

.

And then I did a test with IE10, and I turned off the fiddler with a WebBrowser test time of 186 milliseconds.

Open fiddler with WebBrowser test time incredibly up to 460 milliseconds!

4 Test Conclusion:

1. The browser has a large impact on the test data, 3 of the core, the fastest for WebKit 246, it is possible that this thing has no peripheral program directly using the kernel relationship, he is the fastest, IE10 followed by 400, Firefox slowest to reach more than 2000, but why Firefox speed so slow, this failed to have the answer.

2. Find a variety of materials, found similar to fiddler This tool uses the principle is, launched the Fiddler, the program will be a Microsoft Internet services as a system agent. As a System agent, all HTTP requests from Microsoft Internet Services (WinInet) go through fiddle before reaching the target Web server, and all HTTP responses flow through fiddler before returning to the client. Then this will explain why the fiddler has slowed down after the opening.

This is often said network hijacking, similar tools and httpwach,httpanalyzer, so that this principle of testing data will lead to data deviation from the actual, not recommended!

3. Cannot use Fiddler this principle, because the WebBrowser limited function and the extension, basically cannot take to the har inside data, this scheme will not be feasible!

4. Recommendation: Use BERSERKJS scheme, the current IP is not resolved, need to find a way to solve this problem!

Page front page load performance test analysis of the feasibility scheme of each tool