Efficient Forwarding Rule Management in Software Defined Network via Subnet-Based Pattern Matching

Efficient Forwarding Rule Management in Software Defined Network via Subnet-Based Pattern Matching

연구 분야: Networking

학회: SN Computer Science

Traditional software-defined networks (SDNs) rely on controllers for forwarding decisions, leading to overhead for both switches and controllers, particularly when dealing with first-time flows or limited Ternary Content Addressable Memory (TCAM) on switches. Existing literature explores forwarding rule compression on controllers and offers limited efficiency gains considering storage limitations and extra processing requirements. This work proposes a novel subnetting-based architecture for SDNs inspired by the TCP/IP model. It leverages TCAM’s wildcard matching capabilities to enable prefix-based IP address matching, similar to traditional IP routers. This significantly reduces switch storage requirements and minimizes switch-to-controller communication. The architecture achieves this by grouping flows based on network prefixes and caching frequently accessed flows, leveraging the inherent structure of IP addresses. Simulations across diverse network sizes demonstrate that our approach utilizing compressed forwarding rules with caching significantly outperforms conventional SDN architecture. This is evident in the reduction of both flow table size on switches and controller interactions compared to traditional methods with a high hit ratio of 99.89%.