[Paper Review] Multiple Constrained Routing Algorithms in Large-Scaled Software Defined Networks

Chenyang Xu, Liangde Tao+3 more|arXiv (Cornell University)|Feb 27, 2019

Software-Defined Networks and 5G6 references5 citations

TL;DR

This paper proposes a fast, heuristic, multi-phase routing algorithm for large-scale Software Defined Networks (SDN) that maximizes throughput under joint bandwidth, delay, and hop constraints. The method computes multiple feasible paths per demand, ranks demands using priority rules, and selects paths to optimize residual bandwidth and constraint satisfaction, achieving over 90% of total possible bandwidth in under 10 seconds on large networks with 10,000 nodes.

ABSTRACT

In this paper, we consider the bandwidth-delay-hop constrained routing problem in large-scaled software defined networks. A number of demands, each of which specifies a source vertex and a sink vertex, are required to route in a given network. We are asked to select a subset of demands, and assign a routing path for each selected demand without violating the hop and delay constraints, while assuring that the bandwidth occupied in each edge is not beyond its capacity. The goal is to maximize the throughput (the total bandwidth) of the selected demands. We develop an efficient heuristic algorithm for the problem, which consists of three main steps, namely, computing feasible paths for each demand, sorting the demands with some priority rules, selecting a path for each demand. The algorithm is tested with networks of actual sizes and topologies, generated by Huawei Technologies Company. The experiments show that the proposed approach outperforms existing algorithms both in throughput and in running time. In the experiments, our algorithm achieves more than 90\% of the total bandwidth of the given demands within 10 seconds. Moreover, a large part of our algorithm can run in parallel which largely speeds up the process when using multi-core processors.

Motivation & Objective

• To address the challenge of maximizing throughput in large-scale SDN environments under strict bandwidth, delay, and hop constraints.
• To design a scalable and efficient routing algorithm suitable for real-time network reconfiguration in SDN.
• To outperform existing algorithms in both throughput and running time on realistic, large-scale network topologies.
• To enable parallel execution for improved performance on multi-core architectures.

Proposed method

• The algorithm consists of three main phases: (1) computing multiple feasible paths for each demand that satisfy bandwidth, delay, and hop constraints; (2) sorting demands using priority rules based on bandwidth, hop count, and residual capacity; (3) selecting the optimal path for each demand to maximize total throughput.
• Priority rules are evaluated to determine the best order for processing demands, with Rule 1 (bandwidth-to-hop ratio) showing superior performance.
• Path selection prioritizes routes with higher residual bandwidth and lower criticality to reduce interference with future demands.
• The algorithm leverages the global network view of SDN controllers to compute paths with full knowledge of edge capacities and QoS constraints.
• The method is designed to be highly parallelizable, enabling significant speedup on multi-core processors.
• It uses an extended Dijkstra algorithm to compute feasible paths under delay and hop constraints, ensuring QoS compliance.

Experimental results

Research questions

• RQ1 How can throughput be maximized in large-scale SDN networks under joint bandwidth, delay, and hop constraints?
• RQ2 What priority rule for demand ordering yields the highest throughput in constrained multi-commodity routing?
• RQ3 How does selecting paths based on residual bandwidth compare to selecting based on hop or delay alone?
• RQ4 Can a heuristic algorithm achieve near-optimal throughput with significantly lower running time than existing methods?
• RQ5 To what extent can the algorithm be parallelized to improve performance on modern multi-core systems?

Key findings

• The proposed algorithm achieves an average throughput of 97.71% of the total available bandwidth on medium-scale instances (Instance A), with a standard deviation of 3.29%.
• On large-scale instances (Instance B), the algorithm achieves an average throughput of 92.55%, outperforming all baseline algorithms.
• The algorithm runs in under 10 seconds on average for networks with approximately 10,000 nodes and 40,000 edges, while baseline algorithms take nearly 90 minutes on average.
• Among path selection strategies, choosing the path with the highest residual bandwidth yields the best performance, achieving 97.71% throughput on Instance A.
• The shortest path (least hop) and least delay path strategies yield significantly lower throughput (95.20% and 91.56%, respectively), indicating suboptimal resource utilization.
• The algorithm demonstrates superior stability and robustness compared to MDA, WSP, and SWP, with lower standard deviation in throughput across test instances.