M2: A Multilevel Multi-Field Rule Compression Algorithm for Software-Defined Networks

M2: A Multilevel Multi-Field Rule Compression Algorithm for Software-Defined Networks

연구 분야: Networking

학회: SN Computer Science

The Software-defined networking paradigm recommends using dumb in-network elements and pushes all smartness toward a central entity called a Network Controller. The network controller forces policies in the form of flow-based rules to use the full potential of smartness. However, network elements are equipped with limited high-speed memories, which restricts the extent to which rules can be fine-grained. Rule placement and rule compression techniques seek to reduce the severity of the problem and make room for more network intelligence. This paper introduces M2, a multilevel compression method that recursively groups similar rules based on multiple network packet fields and compresses each group with a wildcard rule. We suggest clear laws to avoid conflicts and discuss the conditions under which the proposed method is effective through extensive simulations of various topologies and rule patterns. The results show a possibility of a 50% compression ratio improvement over its single-level method. M2 is not only effective but also highly practical. It is a lightweight solution that can be applied to large rule tables, creating space for additional rules without incurring significant costs.