Hi Pat et al,
how were the RTT arrived at for the various bandwidth/connection types?
I.e how was 50ms for DSL (1.5 Mbps/384 Kbps) determined?
I know we can the values for our private instances, and after reading Akamai’s state of the internet i was thinking of doing this, but i don’t know how that figure was arrived at.
cheers,
p.s you cant search the forums for the term “rtt”
Largely from the FCC study and updates to it, a bunch of testing that I had done a while back as well as data from several large non-public sources that corroborate the numbers.
Here was the post when we switched to cable: http://www.webpagetest.org/forums/showthread.php?tid=12112
And for the 3G profile: http://www.webpagetest.org/forums/showthread.php?tid=12141
Here is the links to the FCC data: https://sites.google.com/a/webpagetest.org/docs/other-resources/2011-fcc-broadband-data
Ah thanks Pat, thats reasonable.
Its interesting though that your figures are gaining traction in terms of becoming the defacto standard.
In the UK we have had a similar study under taken by OFCOM, and their headlines of the findings are here:
full report here:
Cool, thanks. Frustrating thing is that everyone loves to study and report the bandwidth but hardly anybody reports latency numbers 
As far as web browsing is concerned, anything over 2-3Mbps is no longer bandwidth constrained and over 5Mbps or so shows no changes at all in performance because of all of the round trips. Maybe with SPDY that will start to change but given the amount of 3rd-party content on most sites I think we’re going to be latency constrained for the foreseeable future.