How to calculate the average of TTFB and TTLB
I am doing a web testing using Microsoft WAST. By using this tool, I ran my test script 3-5 times to get the average value of TTFB and TTLB of the response time. However, when I get the values, I am facing a problems to make sure the results are consistent. For example
Test TTFB TTLB
1 23.92 21215.32
2 20.68 21225.04
3 24.4 20823.72
4 21.56 20634.2
5 18.08 20643.44
My doubt is how to determine that results are consistent so that i can calculate for the average for all of the test? What I know from literature, for RTT we have to make sure it consistent with standard deviation less than 1.
However, by using this tool, I think it impossible for me to get the result with stdev less than 1 as to achieve that kind of standard. Is there any theory or standard to determine the consistency of TTFB and TTLB?
Re: How to calculate the average of TTFB and TTLB
Since I haven't used this tool, I have no idea what TTFB and TTLB are. However, your numbers look close enough to me. The first set is higher than the rest and this may be due to caching. If your system makes good use of cache and your data is very close to production data and not unusually repetitive, I might throw out that first number. However, in other circumstances I might only consider the first valid. Assuming your data is good, I would toss the first result and average the other 4. You're always going to have some variation because you can't absolutely control what the OS is going to do in the middle of your test. Longer test runs will of course be more consistent than shorter ones.