ASDO: An Efficient Algorithm for Traffic Engineering in Large-Scale Data Center Network

2025-04-05

AI Summary

Large-scale datacenter network (DCN) traffic engineering (TE) faces scalability bottlenecks: state-of-the-art approaches—whether commercial solver–based or deep learning–based—suffer severe performance degradation or excessive runtime in ultra-large-scale deployments. This paper proposes the Alternating Source-Destination Optimization (ASDO), a novel sequential algorithm for scalable TE. Its core contributions are: (1) the first real-time link utilization–driven dynamic SD-demand scheduling mechanism; and (2) a Balanced Binary Search Method (BBSM) that selects, among multiple optimal flow-splitting ratios, the one minimizing the maximum link utilization for improved load balancing. Evaluated on Meta’s production topology, ASDO reduces normalized maximum link utilization by 65% and 60% compared to TEAL and POP, respectively, while accelerating POP by 12×. These results demonstrate substantial improvements in both efficiency and scalability for large-scale DCN TE.

Technology Category

• Search and Optimization: Algorithm Configuration
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Rapid growth of data center networks (DCNs) poses significant challenges for large-scale traffic engineering (TE). Existing acceleration strategies, which rely on commercial solvers or deep learning, face scalability issues and struggle with degrading performance or long computational time. Unlike existing algorithms adopting parallel strategies, we propose Alternate Source-Destination Optimization (ASDO), a sequential algorithm for TE. ASDO decomposes the problem into subproblems, each focused on adjusting the split ratios for a specific source-destination (SD) demand while keeping others fixed. To enhance the efficiency of subproblem optimization, we design a Balanced Binary Search Method (BBSM), which identifies the most balanced split ratios among multiple solutions that minimize Maximum Link Utilization (MLU). ASDO dynamically updates the sequence of SDs based on real-time utilization, which accelerates convergence and enhances solution quality. We evaluate ASDO on Meta DCNs and two wide-area networks (WANs). In a Meta topology, ASDO achieves a 65% and 60% reduction in normalized MLU compared to TEAL and POP, two state-of-the-art TE acceleration methods, while delivering a $12 imes$ speedup over POP. These results demonstrate the superior performance of ASDO in large-scale TE.

Problem

• Addresses scalability issues in large-scale traffic engineering
• Proposes ASDO algorithm for efficient TE optimization
• Reduces maximum link utilization and computational time

Innovation

• Sequential algorithm for traffic engineering optimization
• Balanced Binary Search Method for efficient subproblem solving
• Dynamic SD sequence updates for faster convergence