BIND: Blockchain-Based Flow-Table Partitioning in Distributed Multi-Tenant Software-Defined Networks

In this paper, we study the problem of flow-table partitioning in distributed multi-tenant software-defined networks (SDNs) having Internet-of-things (IoT) devices. In the existing literature, the optimal usage of the ternary content-addressable memory (TCAM) is studied in the context of data traffic management by introducing the soft flow-table partitioning in the presence of a centralized controller. However, in the presence of distributed multi-tenant controllers, the soft flow-table partitioning may introduce a monopoly among the controllers. Hence, there is a need to design a flow-table partitioning scheme for distributed multi-tenant SDN, while maximizing the network sustainability and throughput. In this work, we propose a utility game-based scheme, named BIND, for dynamic flow-table partitioning. To ensure cooperation among the controllers and to avoid monopoly, we introduce the use of a blockchain among the multi-tenant controllers. Additionally, using BIND, we ensure that each controller gets a fair chance for flow-rule replacement. Moreover, network sustainability is ensured in BIND, while minimizing the flow-rule replacement in the flow-tables and multi-tenant SDN. Through simulation, we observe that using BIND, fairness in flow-rule placement is ensured. Additionally, the network overhead is reduced significantly.