The paper "Wide-Area Traffic: The Failure of Poisson Modeling" by Vern Paxson and Sally Floyd evaluates 21 wide-area traces to investigate the accuracy of modeling network arrivals as Poisson processes. The authors find that while user-initiated TCP session arrivals, such as remote-login and file-transfer, are well-modeled as Poisson processes with fixed hourly rates, other connection arrivals deviate significantly from Poisson. Specifically, TELNET packet interarrivals are underestimating the burstiness of TELNET traffic when modeled as exponentially distributed, but using the empirical Tcplib distribution preserves burstiness over various time scales. FTPDATA connection arrivals within FTP sessions are clustered into "connection bursts," with the largest bursts dominating FTPDATA traffic. The paper also discusses the implications of these findings for congestion control and traffic performance, suggesting that wide-area traffic is more bursty than Poisson models predict. Finally, the authors explore the possibility of self-similarity in wide-area traffic, noting that while the evidence is inconclusive, the traffic exhibits large-scale correlations inconsistent with Poisson processes.The paper "Wide-Area Traffic: The Failure of Poisson Modeling" by Vern Paxson and Sally Floyd evaluates 21 wide-area traces to investigate the accuracy of modeling network arrivals as Poisson processes. The authors find that while user-initiated TCP session arrivals, such as remote-login and file-transfer, are well-modeled as Poisson processes with fixed hourly rates, other connection arrivals deviate significantly from Poisson. Specifically, TELNET packet interarrivals are underestimating the burstiness of TELNET traffic when modeled as exponentially distributed, but using the empirical Tcplib distribution preserves burstiness over various time scales. FTPDATA connection arrivals within FTP sessions are clustered into "connection bursts," with the largest bursts dominating FTPDATA traffic. The paper also discusses the implications of these findings for congestion control and traffic performance, suggesting that wide-area traffic is more bursty than Poisson models predict. Finally, the authors explore the possibility of self-similarity in wide-area traffic, noting that while the evidence is inconclusive, the traffic exhibits large-scale correlations inconsistent with Poisson processes.