Inspired by the following @ericlaw tweet, I tested new Fiddler alpha's ( v2.3.2.0) scalability as it compares to the current version v2.3.1.2.
Session load was created by StresStimulus extension. The following scenario was tested:
- Accessing the Facebook home page (24 requests on empty browser cache),
- with 20 concurrent users, using IE8 with zero think times, until the page is loaded 40x.
I tested this scenario 5 times in a row with each of the two Fiddler versions. The average of 5 are below:
| Load Test Results |
Current v2.3. 1.2 |
Alpha v2.3. 2.0 |
Improvement |
| Test run duration (s) |
16.748 |
12.914 |
29.7% |
| Total requests |
960 |
960 |
|
| Requests/sec |
57.484 |
76.283 |
32.7% |
| Avg. iteration time (s) |
7.306 |
5.594 |
30.6% |
| Avg. response time (s) |
0.1074 |
0.1024 |
4.7% |
New Alpha improved the load test metrics by about 30%.
Brief discussion of the results
The ideal stress test of Fiddler scalability should be conducted in the settings where the Fiddler, by itself, is the only major bottleneck. In this ideal case the 30% of Fiddler performance increase would lead to about 30% improvement of the test metrics. In my case test Fiddler was not the only bottleneck. While my Internet connection speed is good, and the web page was fast and relatively small, it was not as fast as it would be on a local dedicated Web server.
In my test in order to achieve 30% improvement of the load test metrics, Fiddler scalability should have been improved more significantly, probably as much as 50% or more. This is great news for Fiddler users!
If you will replicate this load test (it took me about 10 minutes to prepare and run all 10 tests), please comment your results.
Note: For testing I used just released StresStimulus version 0.8.4043 that has a number of fixes.