WebPagetest Forums
Best way to get the best results - Printable Version

+- WebPagetest Forums (https://www.webpagetest.org/forums)
+-- Forum: Web Performance (/forumdisplay.php?fid=3)
+--- Forum: Optimization Discussions (/forumdisplay.php?fid=5)
+--- Thread: Best way to get the best results (/showthread.php?tid=323)



Best way to get the best results - gokninski - 07-29-2010 02:37 AM

Hello there,

I have a bit of a general question that I would like to get some opinions on.

I am testing a website that is located in the UK but is accessed by customers from the UK, US, CA, AUS as well as countires in Asia and I am also wanting to get the most accurate results with the view to then carry out the tests again following some optimisation work.

So, what I would like to know is what is the best/most accurate way to measure an internationally accessed website? Bearing in mind the different locations and potentially different levels of traffic.

It would be awesome if I was able to carry out a test, and then carrying it out again a few days later, I would get similiar results. But how can I ensure this?

I apologise if my quesiton is too general or maybe even if it is not completely possible but your opinions are appreciated.

Thanks in advance,

Greg


RE: Best way to get the best results - pmeenan - 07-29-2010 03:02 AM

You really only have 2 options:

1 - Measure your actual user performance (using something like boomerang - http://developer.yahoo.net/blog/archives/2010/06/performance_testing_with_boomerang.html )

2 - Run actual tests from the various locations you are interested in. WebPagetest has some of the locations covered but not all. You need to be careful to use reasonable consumer connectivity when testing (so be careful using a service that has test machines connected to the backbone). Make sure to run several tests (at least 10) to get a solid baseline from each location and eliminate any outliers.

Measuring the field performance is the best way to gauge what your customers are actually seeing but the results can vary quite a bit and you don't have any control over the conditions. The synthetic testing will produce more consistent results for comparison but may not represent exactly what your user base is seeing.

-Pat


RE: Best way to get the best results - gokninski - 07-29-2010 03:12 AM

(07-29-2010 03:02 AM)pmeenan Wrote:  You really only have 2 options:

1 - Measure your actual user performance (using something like boomerang - http://developer.yahoo.net/blog/archives/2010/06/performance_testing_with_boomerang.html )

2 - Run actual tests from the various locations you are interested in. WebPagetest has some of the locations covered but not all. You need to be careful to use reasonable consumer connectivity when testing (so be careful using a service that has test machines connected to the backbone). Make sure to run several tests (at least 10) to get a solid baseline from each location and eliminate any outliers.

Measuring the field performance is the best way to gauge what your customers are actually seeing but the results can vary quite a bit and you don't have any control over the conditions. The synthetic testing will produce more consistent results for comparison but may not represent exactly what your user base is seeing.

-Pat

Fantastic, thank you so much for your prompt and helpful reply!