• •    

DPCS2017+71+面向大规模数据接入系统的负载平衡机制

周岳1,陈庆奎2   

  1. 1. 上海理工大学光电信息工程学院
    2. 上海理工大学 光电信息与计算机工程学院,上海 200093;
  • 收稿日期:2017-07-25 修回日期:2017-08-03 发布日期:2017-08-03
  • 通讯作者: 周岳

DPCS2017+71+Load balancing mechanism for large-scale communication system

yue zhou,CHEN Qingkui   

  • Received:2017-07-25 Revised:2017-08-03 Online:2017-08-03
  • Contact: yue zhou

摘要: 针对当前分布式系统负载平衡算法存在:(1)算法建立的系统中各节点角色固定,系统不具有自适应性。(2)算法的通用性不高。(3)负载迁移任务巨大,且负载平衡周期过长等问题。本文提出了混合式负载平衡算法。首先,本文设计了一个分布式系统接收模型。模型将系统任务分为三层:接收层、处理层和存储层。在接收层本文使用了自定义的通信协议提高系统的接收性能。然后,负载平衡算法采用随机负载迁移策略,根据系统中节点的负载状态,负载任务进行随机迁移。通过这种策略解决负载周期过长和负载回迁问题。最后,通过分布式控制节点选择策略,使系统中节点具有自适应性。实验结果显示,在百万数据源以下,系统各层平均延迟处于毫秒级,系统负载平衡平均耗时在3分钟以下。实验证明了,本文提出的负载平衡机制具有周期短,任务响应迅速等特点,能够提高分布式系统的接收性能。

关键词: 大规模接入系统, 分布式系统, 混合式负载平衡, 随机负载迁移, 数据接入系统

Abstract: To solve some problems of the current load balancing algorithms for distributed systems:(1) The role of each node in the system is fixed, the system has no adaptability. (2) Load balancing algorithm is not universal. (3) The migration task is too large, and the load cycle is too long. A hybrid load balancing algorithm was presented by this paper. Firstly, a distributed receiving system model was designed by this paper. Model was divided into three parts: receiving level,handling level and saving level. In receiving level, a home-made transmission protocol was used to improve the reception capability of the system. And then, the random load transfer strategy was used by load balancing algorithm. According to the status of the node, the task of load was randomly transferred. the problems of long load cycle and load moving back were solved by this strategy. Finally, distributed control node selecting strategy was adopted to give the nodes the adaptability capability. The experimental results show that the average delay in each layer of the system is in milliseconds, and the system load balancing takes less than 3 minutes. Experiment prove that the load balancing mechanism has short load cycle and fast response, etc. It could improve the reception capability of the distributed system.

Key words: Large-scale Communication System, Distributed Parallel System, Hybrid Load Balance, Random Load Migration, Data Access System

中图分类号: