Catching popular prefixes at AS border routers with a prediction based method

  • Authors:
  • Wei Zhang;Jun Bi;Jianping Wu;Baobao Zhang

  • Affiliations:
  • Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China and National Academy of Defense Information, Wuhan 430010, China;Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China and The Network Research Center, Tsinghua University, Beijing 100084, China;Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China and The Network Research Center, Tsinghua University, Beijing 100084, China;Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China

  • Venue:
  • Computer Networks: The International Journal of Computer and Telecommunications Networking
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Modern Internet routers require powerful forwarding facilities to cope with extremely high rate Forwarding Information Base (FIB) lookups. In general, the FIB is constrained to a small highly efficient but expensive memory. Unfortunately, the BGP route table (RIB) keeps increasing, and this subsequently results in severe FIB inflation at BGP routers. What if we only load a small portion of the RIB into the FIB? Recently the route caching mechanism has been revisited. With such a route caching mechanism, the optimal method is to load in a FIB with popular prefixes which contribute major traffic loads. We propose a prediction based method to catch those popular prefixes with a limited cache size. In this paper, the dynamics of popular prefixes has been studied based on real traffic traces from different ISPs. On applying a GM(1,1) model which is widely applied in grey system control and prediction, we propose a traffic prediction-based route caching method which attempts to bias the cache dump strategy with a range of history to ameliorate the effects of bursts from non-popular prefixes. We also suggest applying FIB aggregation techniques, e.g. Optimal Routing Table Constructor (ORTC) algorithm, to suppress the number of non-popular sub-prefixes of the popular prefixes on route updates. The evaluation of our method is based on simulation over real traffic traces. The simulation shows our prediction-based cache replacement strategy outperforms other cache strategies and matches Internet traffic dynamics very well.