Abstract

Predicting the throughput of large TCP transfers is important for a broad class of applications. This paper focuses on the design, empirical evaluation, and analysis of TCP throughput predictors. We first classify TCP throughput prediction techniques into two categories: Formula-Based (FB) and History-Based (HB). Within each class, we develop representative prediction algorithms, which we then evaluate empirically over the RON testbed. FB prediction relies on mathematical models that express the TCP throughput as a function of the characteristics of the underlying network path. It does not rely on previous TCP transfers in the given path, and it can be performed with non-intrusive network measurements. We show, however, that the FB method is accurate only if the TCP transfer is window-limited to the point that it does not saturate the underlying path, and explain the main causes of the prediction errors. HB techniques predict the throughput of TCP flows from a time series of previous TCP throughput measurements on the same path, when such a history is available. We show that even simple HB predictors, such as Moving Average and Holt-Winters, using a history of few and sporadic samples, can be quite accurate. On the negative side, HB predictors are highly path-dependent. We explain the cause of such path dependencies based on two key factors: the load on the path and the degree of statistical multiplexing.

Citations

988	Modeling tcp throughput: a simple model and its empirical validation - Padhye, Firoiu, et al. - 1998
854	Resilient overlay networks - Andersen, Balakrishnan, et al. - 2001
631	Equation-based congestion control for unicast applications - Floyd, Handley, et al. - 2000
463	The Macrosopic Behavior of the TCP Congestion Avoidance Algorithm - MATHIS, SEMKE, et al. - 1997
268	Enabling Conferencing Applications on the Internet using an Overlay Multicast Architecture - Chu, Rao, et al. - 2001
257	End-to-end available bandwidth: measurement methodology, dynamics, and relation with TCP throughput - Jain, Dovrolis - 2003
250	Topologically-aware overlay construction and server selection - Ratnasamy, Handly, et al. - 2002
178	pathChirp: Efficient available bandwidth estimation for network paths - Ribeiro, Riedi, et al. - 2003
170	Modeling TCP Latency - CARDWELL, SAVAGE, et al. - 2000
150	Statistical Bandwidth Sharing: A Study of Congestion at Flow Level - Fred, Bonald, et al. - 2001
79	A comparison of overlay routing and multihoming route control - Akella, Pang, et al. - 2004
58	Predicting the performance of wide-area data transfers - Vazhkudai, Schopf, et al. - 2002
53	A predictability analysis of network traffic - Sang, Li - 2000
32	Predicting TCP Throughput From Non-invasive Network Sampling - Goyal, Guérin, et al. - 2002
26	2002): Multivariate resource performance forecasting in the network weather service - Swany, Wolski
23	Foundations of Time Series Analysis and Prediction Theory - Pourahmadi - 2001
14	A Unified Framework for Understanding Network Traffic Using Independent Wavelet Models - Tian, Wu, et al. - 2002
10	An Empirical Study of the Multiscale Predictability of Network Traffic - Qiao, Skicewicz, et al. - 2004
3	Variance Changes in Time Series - Outliers - 1988
3	An Integrated Model for the Latency and Steady-State - Sikdar, Kalyanaraman, et al. - 2001