Understanding BGP Session Robustness in Bandwidth Saturation Regime

The reliability and robustness of the Border Gateway Protocol (BGP) play very important roles in achieving highly stable and prompt Internet data communication. The present BGP uses TCP/IP to exchange routing control information. These routing control messages are usually not differentiated from the normal data packets in Internet packet forwarding, and this makes BGP sensitive to severe network congestion. In this paper, we investigate the packet drop probability and the lifetime of BGP sessions in two types of bandwidth saturation scenarios: (1) the TCP bandwidth saturation caused by traffic engineering failures; (2) the UDP bandwidth saturation caused by worm attacks. In the TCP bandwidth saturation, as time elapses, the packet drop probability decreases in a subexponential way. While, in the UDP bandwidth saturation, the packet drop probability converges in an exponential way to its maximum value with the progress of worm propagation. For the lifetime of BGP sessions, based on our empirical study, we find that in most cases the BGP session lifetime can be characterized using exponential distributions and Weibull distributions. In the case of TCP bandwidth saturation, if all TCP connections have the same round trip time, the tail of the BGP lifetime tends to be power-law. To get a deep understanding of the BGP sessions lifetime, we give an approximate model for the expected lifetime of BGP sessions, and show that by slightly changing the TCP retransmission parameters, the expected BGP session lifetime can be improved significantly. Our research results provide very important index in evaluating the reliability of Internet routing schemes and are very helpful in designing resilient Internet routing infrastructures.

[1]  Aman Shaikh,et al.  Routing stability in congested networks: experimentation and analysis , 2000 .

[2]  David Watson,et al.  An extensible probe architecture for network protocol performance measurement , 2004, Softw. Pract. Exp..

[3]  Kishor S. Trivedi Probability and Statistics with Reliability, Queuing, and Computer Science Applications , 1984 .

[4]  Norman T. J. Bailey,et al.  The Mathematical Theory of Infectious Diseases , 1975 .

[5]  Armand M. Makowski,et al.  Limit behavior of ECN/RED gateways under a large number of TCP flows , 2003, IEEE INFOCOM 2003. Twenty-second Annual Joint Conference of the IEEE Computer and Communications Societies (IEEE Cat. No.03CH37428).

[6]  Aman Shaikh,et al.  Routing stability in congested networks: experimentation and analysis , 2000, SIGCOMM 2000.

[7]  Daniel Massey,et al.  Observation and analysis of BGP behavior under stress , 2002, IMW '02.

[8]  Donald F. Towsley,et al.  Modeling TCP Reno performance: a simple model and its empirical validation , 2000, TNET.

[9]  Susan Hares,et al.  A Border Gateway Protocol 4 (BGP-4) , 1994, RFC.

[10]  Yakov Rekhter,et al.  A Border Gateway Protocol 4 (BGP-4) , 1994, RFC.

[11]  Farnam Jahanian,et al.  Internet routing instability , 1997, SIGCOMM '97.

[12]  J. Klein,et al.  Survival Analysis: Techniques for Censored and Truncated Data , 1997 .

[13]  Klara Nahrstedt,et al.  Reliability models and evaluation of internal BGP networks , 2004, IEEE INFOCOM 2004.

[14]  W. Richard Stevens,et al.  TCP/IP Illustrated, Volume 2: The Implementation , 1995 .