Detecting and Profiling TCP Connections Experiencing Abnormal Performance
The authors study functionally correct TCP connections - normal set-up, data transfer and tear-down - that experience lower than normal performance in terms of delay and throughput. Several factors, including packet loss or application behavior, may lead to such abnormal performance. They present a methodology to detect TCP connections with such abnormal performance from packet traces recorded at a single vantage point. Their technique decomposes a TCP transfer into periods where: TCP is recovering from losses, the client or the server are thinking or preparing data, respectively, or the data is sent but at an abnormally low rate.