摘要
分布式计算集群可以很好地解决计算性能瓶颈和横向扩展的成本问题,但实际工程中同时会带来合理调度和充分利用资源的新需求。通过远程方法调用的方式实现主从架构的任务调度系统,同时结合基于多参数的负载均衡算法和Java的多线程机制,提高系统的资源使用率和稳定性,满足生产时的复杂任务编排和计算密集型任务的性能需求。
Distributed computing clusters can solve the problems of computational performance bottlenecks and horizontal scaling,but in actual engineering,new requirements for rational scheduling and full utilization of resources will be brought about.In this paper,we design the task scheduling platform of the master-slave architecture is realized by remote method invocation.At the same time,the multi-parameters based load balancing algorithm and Java multi-threading mechanism are combined to improve resource utilization and stability.The platform is also designed to meet the complex task scheduling and production with performance requirements for computeintensive tasks.
作者
张楠
李宗骍
Zhang Nan;Li Zong-xing
出处
《电力系统装备》
2018年第10期216-217,共2页
Electric Power System Equipment
关键词
高性能
分布式计算
任务调度
负载均衡
High performance
Distributed Computing
Task Scheduling
Load Balancing
