We have seen in a previous post that it was possible to measure the load time of web pages from a web console. The data collection is fast and accurate, but we stay dependent on the network where we are localized. It can be interesting to measure a load time of web page around the globe
Indeed the measures from the web console are correct only for your environment/web connection. What’s the measure in another place of the world?
Is there an impact? I would like to ask “Yes”. Now we will try to prove it.
When we write a website address, we connect to a web server linked to an address and a content.
If our web page is hosted in France and that we are in the USA, information will have to cross the ocean through a cable underwater, just like a freeway connects several towns. Let’s measure the load time of a web page with online tools.
For my test, we will use « LyonTesting » website and following tools for measurements
There are probably other tools, please share those you know.
I will present WebpageTest more in another article, because this tool can be very complete for performance measures.
The website in hosted by OVH (Whois) and I can measure the response time of the home page with the the web console (with Chrome 56)
All these websites use the same method to measure the load time.
We copy/paste the link, select the server and the browser, then the search is launched after the end of last search (not to falsify the measure)
Example with Pingdom
Server locate in the same place (or nearest)
- Pingdom – Région “New York City, USA” – Chrome 39
- WebpageTest – Région “Virginie, USA” – Chrome 56
- Gtmetrix – Région “Vancouver, Canada” hors protocole – Firefox 49
We see with these results that for “WebpageTest”, we are closed to values returned by the web console.
However the site “Pingdom” return a faster measurement compared to the web console.
We can think that a part of the page isn’t loading but that’s not true. The size of the page (2.1MB) is loaded and the amount of request is the same (console: 67, Pingdom: 70, WebpageTest: 66 Gtmetrix: 62). Strange no?
Is Chrome 39 faster than Chrome 56? The results makes us think that it is. Idea to explore
Server located in another place (other continent)
- Pingdom Région “Melbourne, Australia” – Chrome 39
- WebpageTest Région “Buenos Aires, Argentina” – Chrome 56
- Gtmetrix “Vancouver, Canada” Hors protocole – Firefox 49
Amazing no ? All the websites took more time for being loaded (around 5s more for Australia)
I then restart the same test for WebpageTest
- WebpageTest Région “Australia” – Chrome 56
I confirm the last result.
These results are correct, we can imagine that the time to contact the server in France takes more that USA. The next question is in Europe?
Server in Europe
- Pingdom Région “Stockholm, Sweden” – Chrome 39
- WebpageTest Région “Brussels, Belgium” – Chrome 56
Whereas the results for Pingdom are similar to the measure in Virginia, we will see that WebPagetest results are closed to the web console results. The response time are lower than from USA. Closer to the server, therefore less distance to travel.
We supposed that it exists an impact if we measure the load time of webpage in another continent.
We widely answer at this supposition because the more the distance is between the host server and the measured web page, the more the load time of the web page is. We also highlighted the importance of the site on which the measurements were made indeed Pingdom and WebpageTest did not give the same results at all.
What is the best value?
Allow me to say that it is “WebpageTest” because it matches with Chrome browser, but more investigations should be done.