On the relationship between fundamental measurements in TCP flows

Abstract

Abstract-This paper considers fundamental measurements which drive TCP flows: throughput, RTT and loss. It is clear that throughput is, in some sense, a function of both RTT and loss. In their seminal paper Padyhe et al [1] begin with a mathematical model of the TCP sliding window evolution process and come up with an equation showing that TCP throughput is (roughly) proportional to 1/RT T √ p where p is the probability of packet loss. Their equation is shown to be consistent with data gathered on several links. This paper takes the opposite approach and analyses a large number of packet traces from well-known sources in order to create a data-driven estimate of the functions which relate TCP, loss and RTT. Regression analysis is used to fit models to connect the quantities. The fitted models show different behaviour from that expected in [1].

On rate limitation mechanisms for TCP throughput: A longitudinal analysis

“…For medium sized flows 1MB to 20MB, there is no clear correlation between flow size and flow rate. This is an unexpected finding and different to findings from previous research [38,18]. Above 20MB the sampling problem because of the 15 minute samples becomes a problem and our data cannot tell us anything about the relationship between flow length and throughput.…”

“…A commonly held assumption is that throughput is correlated with flow size, which has been verified empirically in previous studies [38,18]. Much of the data used in these studies however precedes the widespread adoption of high bandwidth connections and use of streaming media, both of which can impact the extent to which contention occurs in the network.…”

A longitudinal analysis of Internet rate limitations

“…Given host limitations are most likely to affect large flows, it's worth considering whether other constraints are applied disproportionately across flow sizes. A commonly held assumption is that throughput is correlated with flow size, which has been verified empirically in previous studies [9], [24]. Much of the data used in these studies however precedes the widespread adoption of high bandwidth connections and use of streaming media, both of which can impact the extent to which contention occurs in the network.…”