Early work in the late 60s and in the 70s [96,10,164,97,175,95] studied the response time of packet and circuit switching in the context of radio networks, satellite communications and the ARPANET (the precursor of the modern Internet). These three network scenarios had something in common: links had less capacity than the bandwidth that an end host could process, and so a single end host could take all link bandwidth along a path if no one else was using it. The conclusion of this early work was that packet switching is more efficient than circuit switching, and it provides better response time under these scenarios. These results were obtained using M/M/N queueing models, where arrivals are Poisson and service times are exponential.
With time, these results have been extrapolated to form part of the IP folklore despite the fact that much has changed in the Internet. First, a single end host is no longer capable of filling up a link in the core (2.5 Gbit/s and above) on its own. Second, it has been shown that whereas flow/session arrivals are Poisson (or close to Poisson) [78,45], flow sizes are not exponential, but rather heavy-tailed, and thus they are closer to a Pareto distribution than an exponential one [84,57,183]. This chapter evaluates the end-user response time with consideration of the characteristics of the current Internet.