On Fri, 3 Sep 2010, Marshall Eubanks wrote: > > On Sep 2, 2010, at 8:45 PM, Phillip Hallam-Baker wrote: > > > So in my view the problem here is that when I pay for an X Mb/sec > > connection at the moment I have no real way of knowing whether that is > > really X Mb/sec all the time or X/n Mb/sec when I am using a service > > that competes with my carrier. > > This sounds like there is potential for crowd sourcing here. > > For example, I can tell you nothing about Vonage, but a fair amount > about Cox Cable Internet. What you want to know is known, just not (yet) > in a way you can easily access. > > Would a Yelp type model be appropriate ? It might tell you something about customer service and might tell you if there is a pattern of promising more than is delivered, but the last mile of the connection is highly variable. Depending on the carrier technologies, there may be distance issues and/or issues re. folks the link is shared with, etc. The only way to know is to make parallel installations and test them using carefully constructed methodologies to try to factor out other variables like origin server load, backbone load, etc. Then there is all the automatic stuff most users aren't aware of which uses up bandwidth and is almost impossible to identify. Close to a no win situation for consumer class services. I know from repeated experience that my installed speed will be downrated from the promised speed. It has never not happened with multiple providers and locations. _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf