Reducing Flow Table Update Costs in Software-Defined Networking.

Area of Science:

• Computer Science
• Networking

Background:

• Software-defined networking (SDN) relies on efficient flow table updates for traffic forwarding.
• Fine-grained control in SDN increases rule insertions/deletions, leading to complex dependencies.
• Overlapping rules in ternary content-addressable memory (TCAM) flow tables necessitate ordered updates, causing latency.

Purpose of the Study:

• To reduce the delay in updating dependent rules within TCAM-based flow tables.
• To address the NP-hard problem of TCAM maintenance and improve forwarding efficiency.
• To propose an algorithm that optimizes rule updates in SDN environments.

Main Methods:

• Formulated TCAM maintenance as an NP-hard problem.
• Proposed an optimal moving chain algorithm for single rule updates with theoretical proof.
• Developed a dynamic approach for heuristically updating multiple concurrent entries within a restricted TCAM region.

Main Results:

• The proposed algorithm achieves approximately 6% fewer moving steps compared to existing methods.
• The efficiency gain is more significant with heavily utilized flow tables and longer rule dependency chains.
• Demonstrated improved update efficiency under varying dependency complexities.

Conclusions:

• The developed algorithm effectively reduces TCAM update latency in SDN.
• Optimized rule management in TCAM flow tables enhances overall network forwarding performance.
• This approach offers a practical solution for managing complex rule dependencies in high-performance networks.