摘要
针对当前突发大数据流在单路径调度中产生的链路拥塞问题,为了减少网络链路拥塞、提高网络性能,提出了一种基于队列管理的媒体优化流量自适应调度算法。首先,制定多策略调度策略,对于用于数据平面上的流列队使用特定选择策略的固定优先级调度,通过软件定义网络(SDN)传输媒体优化流量。然后对不同调度策略和条件下SDN中资源管理的有效性进行了全面分析,通过对单个应用资源分配问题和全网应用资源分配问题的迭代求解,得出优化问题的近似最优解。最后与传统的MPLSTE和ECMP算法进行性能对比和分析。结果表明:提出的自适应调度算法本文方法比MPLSTE和ECMP算法能够多转发约9%的流量,在平均链路利用率和平均传输时延几项性能指标上,均优于对比算法。
Aiming at the link congestion problem caused by the current burst large data stream in single path scheduling,in order to reduce network link congestion and improve network performance,proposed a media optimized traffic adaptive scheduling algorithm based on queue management. First,a multi-strategy scheduling policy is developed to optimize traffic for media-defined network( SDN)transmission media for fixed-priority scheduling using a particular selection policy for flow queues on the data plane. Then,the effectiveness of resource management in SDN under different scheduling strategies and conditions is comprehensively analyzed. By solving the problem of single application resource allocation and the application of resource allocation in the whole network,the approximate optimal solution of the optimization problem is obtained. Finally,compared with the traditional MPLSTE and ECMP algorithms,the results show that the proposed adaptive scheduling algorithm can forward about 9% more traffic than the MPLSTE and ECMP algorithms,with average link utilization and average transmission. Several performance indicators are better than the comparison algorithm.
作者
汪海涛
余松森
WANG Haitao;YU Songsen(College of Information&Automation,Guangdong Polytechnic of Science and Trade,Guangzhou 510430,China;School of Software,South China Normal University,Foshan 528225,China)
出处
《重庆理工大学学报（自然科学）》
CAS
北大核心
2020年第7期176-184,共9页
Journal of Chongqing University of Technology：Natural Science
基金
国家自然科学基金项目(61572028)
广东省普通高校特色创新类项目(2019GKTSCX038)
广东省教育厅与思科公司产学合作协同育人项目-高职计算机类专业综合改革项目(粤教职函[2018]182号)。
关键词
链路拥塞
调度策略
网络资源异构性
资源分配
迭代求解
link congestion
scheduling strategy
network resource heterogeneity
resource allocation
iterative solution
