WebPagetest Forums
RTT vs. Dummynet Traffic throttling - Printable Version

+- WebPagetest Forums (https://www.webpagetest.org/forums)
+-- Forum: WebPagetest (/forumdisplay.php?fid=7)
+--- Forum: General Discussion (/forumdisplay.php?fid=25)
+--- Thread: RTT vs. Dummynet Traffic throttling (/showthread.php?tid=13662)



RTT vs. Dummynet Traffic throttling - Christian - 04-10-2015 01:21 AM

Hello,

I would like to get a better understanding of how the traffic WPT.org traffic RTT setting works and what I need to consider during my analyzes.

AFAIK WPT.org does use the dummynet solution to
a) shape the bw and
b) the delay option to influence the RTT.

Therefore I have to believe that the RTT mentioned inside the connectivity.ini files for the different connections is something I need to consider to add on top of the additional network traffic latency introduced by the real network and all kind of other included devices right?

If this is right I would need to calculate the real RTT as follow:
real RTT = RTT - delay.
A good way to measure the RTT is to use a tcp dump and lookup the initial tcp handshake.
If i do so for the russia/moscow agent with the Connection "DSL RTT=50ms" i do get the following values:
49ms = 96ms - 50ms

I did test en.wikipedia.org (91.198.174.192) from the Moscow Agent (188.120.233.22) which does show an RTT via the Traceroute of 44ms.

This fits very well into my assumption, still i wanted to ask if this is the correct way of thinking.

Thanks
Christian


RE: RTT vs. Dummynet Traffic throttling - pmeenan - 04-10-2015 02:47 AM

Sort of. The RTT in connectivity is supposed to be used for adding last-mile latency to the RTT, not for controlling the full RTT to the various servers.

If you select "Native" from the connection type you can test with no added latency but yes, your assumptions and way of working back to the real network RTT are correct.