Minimization of TCAM Usage for SDN Scalability in Wireless Data Centers

The software-defined networking (SDN) architecture has been widely deployed in data centers to support diverse cloud services. In order to provide fast access for flow table lookup, SDN switches usually use ternary content addressable memory (TCAM) to store their routing rules given by a controller. However, the number of TCAM entries is relatively small due to its high cost and high energy consumption. As traffic in data centers significantly increases, rule replacement will occur very frequently in SDN switches. As a result, massive flow setup requests are generated and cannot be processed by the controller in real time. To address the scalability issue, we propose a novel rule-reduction-based policy to minimize TCAM usage, thus alleviating the burden on the SDN controller. Specifically, we integrate the 60 GHz wireless technology into data centers and formulate a routing rule minimization problem under the hybrid architecture with wired and wireless links. The objective is to minimize the number of installed rules in switches. We then propose an efficient algorithm, with the consideration of TCAM limitations and flow sizes, to solve the problem. Results of a series of simulations show that our algorithm outperforms a well-known solution for wired data centers and two approaches for wireless data centers. The results also provide useful insights and guidelines for designing scalable SDN-based data centers.