Efficient TCAM Rules Distribution Algorithms in Software-Defined Networking

Abstract

In software-defined networking (SDN), network rules are installed in ternary content-addressable memory (TCAM). TCAM is a scarce and expensive resource which is a bottleneck for scaling SDN. Rule distribution is a strategy to solve the TCAM shortage problem, which decomposes a large table stored at network ingress into a number of smaller sub-tables and distributes them across network switches. Rule distribution includes two sub-problems, sub-table allocation, and table decomposition. Sub-table allocation is to guarantee that each flow path passes all the partitioned sub-tables and maximize the number of partitioned sub-tables (for reducing the size of sub-tables). Table decomposition problem is to partition a large rule table into a number of balanced sub-tables and reduce the rules overhead. In this paper, we propose a sub-table allocation algorithm and size-balancing sub-table partition algorithm. Simulation results show that the performance of both algorithms is better than previous works in terms of reducing the TCAM entries used in each switch.