A versatile model for TCP bandwidth sharing in networks with heterogeneous users

Enabled by the emergence of various access technologies (such as ADSL and wireless LAN), the number of users with high-speed access to the Internet is growing rapidly, and their expectation with respect to the quality-of-service of the applications has been increasing accordingly. With TCP being the ubiquitous underlying end-to-end control, this motivates the interest in easy-to-evaluate, yet accurate, performance models for a TCP-based network shared by multiple classes of users. Building on the vast body of existing models, we develop a novel versatile model that explicitly captures user heterogeneity, and takes into consideration dynamics at both the packet level and the flow level. It is described as to how the resulting multiple time-scale model can be numerically evaluated. Validation is done by using NS2 simulations as a benchmark. In extensive numerical experiments, we study the impact of heterogeneity in the round-trip times on user-level characteristics such as throughputs and flow transmission times, thus quantifying the resulting bias. We also investigate to what extent this bias is affected by the networks’ ‘packet-level parameters’, such as buffer sizes. We conclude by extending the single-link model in a straightforward way to a general network setting. Also in this network setting the impact of heterogeneity in round-trip times is numerically assessed.

[1]  François Baccelli,et al.  Flow level simulation of large IP networks , 2003, IEEE INFOCOM 2003. Twenty-second Annual Joint Conference of the IEEE Computer and Communications Societies (IEEE Cat. No.03CH37428).

[2]  J. W. Cohen,et al.  The multiple phase service network with generalized processor sharing , 1979, Acta Informatica.

[3]  Alexandre Proutière,et al.  Statistical bandwidth sharing: a study of congestion at flow level , 2001, SIGCOMM.

[4]  Vishal Misra,et al.  Fluid-based analysis of a network of AQM routers supporting TCP flows with an application to RED , 2000, SIGCOMM.

[5]  Donald F. Towsley,et al.  Modeling TCP throughput: a simple model and its empirical validation , 1998, SIGCOMM '98.

[6]  Thomas Bonald,et al.  Statistical bandwidth sharing: a study of congestion at flow level , 2001, SIGCOMM.

[7]  T. V. Lakshman,et al.  The performance of TCP/IP for networks with high bandwidth-delay products and random loss , 1997, TNET.

[8]  Pirkko Kuusela,et al.  Modeling RED with Idealized TCP Sources , 2001 .

[9]  Matthew Mathis,et al.  The macroscopic behavior of the TCP congestion avoidance algorithm , 1997, CCRV.

[10]  Isi Mitrani,et al.  Sharing a Processor Among Many Job Classes , 1980, JACM.

[11]  Donald F. Towsley,et al.  Fixed point approximations for TCP behavior in an AQM network , 2001, SIGMETRICS '01.

[12]  F. Kelly.  Blocking probabilities in large circuit-switched networks , 1986, Advances in Applied Probability.

[13]  Eitan Altman,et al.  Fairness analysis of TCP/IP , 2000, Proceedings of the 39th IEEE Conference on Decision and Control (Cat. No.00CH37187).

[14]  Laurent Massoulié,et al.  Arguments in favour of admission control for TCP flows , 1999 .

[15]  Richard J. Gibbens,et al.  Fixed-Point Models for the End-to-End Performance Analysis of IP Networks , 2000 .

[16]  Anurag Kumar,et al.  Stochastic models for throughput analysis of randomly arriving elastic flows in the Internet , 2002, Proceedings.Twenty-First Annual Joint Conference of the IEEE Computer and Communications Societies.

[17]  Ugo Mocci,et al.  Broadband Network Teletraffic: Final Report of Action COST 242 , 1996 .

[18]  Michel Mandjes,et al.  Performance evaluation of strategies for integration of elastic and stream traffic , 1999 .

[19]  Michel Mandjes,et al.  An integrated packet/flow model for TCP performance analysis , 2003 .

[20]  Patrick Brown,et al.  Resource sharing of TCP connections with different round trip times , 2000, Proceedings IEEE INFOCOM 2000. Conference on Computer Communications. Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies (Cat. No.00CH37064).

[21]  Eitan Altman,et al.  DPS queues with stationary ergodic service times and the performance of TCP in overload , 2004, IEEE INFOCOM 2004.

[22]  Eitan Altman,et al.  The effect of router buffer size on the TCP performance , 2001 .

[23]  Hyoung-Kee Choi,et al.  A behavioral model of Web traffic , 1999, Proceedings. Seventh International Conference on Network Protocols.

[24]  Nicky van Foreest,et al.  A versatile model for asymmetric TCP sources , 2003 .

[25]  Boudewijn R. Haverkort,et al.  Versatile Markovian models for networks with asymmetric TCP sources , 2004 .

[26]  Laurent Massoulié,et al.  Impact of fairness on Internet performance , 2001, SIGMETRICS '01.