Space and speed tradeoffs in TCAM hierarchical packet classification

  • Authors:
  • Alexander Kesselman;Kirill Kogan;Sergey Nemzer;Michael Segal

  • Affiliations:
  • Google Inc., United States;Computer Science Department, University of Waterloo, Canada;Compugen, Israel;Department of Communication Systems Engineering, Ben-Gurion University, Israel

  • Venue:
  • Journal of Computer and System Sciences
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

Traffic classification in the Internet is a crucial mechanism necessary to support network services. Using Ternary Content-Addressable Memories (TCAMs) to perform high-speed packet classification has become the de facto standard in industry. TCAMs concurrently match the packet headers against the rules in a classification database providing high throughput unparalleled by software-based solutions. The complexity of packet classification policies has been growing rapidly as the number of Internet services continues to increase. Many complex classification policies are naturally represented in a hierarchical fashion, where different layers perform classification based on the administrative domain and the traffic QoS parameters. However, multiple levels of classification hierarchy incur high lookup latency while high TCAM memory requirements of flattened classification policies is a major issue since TCAMs have very limited capacity. In this paper we focus on the fundamental tradeoff between the TCAM space and the number of lookups in the TCAM classification policies. We consider two optimization problems of dual nature: the first problem is to minimize the number of TCAM entries subject to the constraint on the maximum number of levels in the policy hierarchy; the second problem is to minimize the number of levels in the policy hierarchy subject to the constraint on the maximum number of TCAM entries. We propose efficient algorithms for these problems, which do not require any hardware changes. To the best of our knowledge, this is the first work to study these problems. We also show experimental results that support our findings.