基于Socket分布式计算的泥石流危险性分区系统

Debris-flow risky district decision system based on socket distributed-computing

针对以往单机版的泥石流分区系统计算速度慢、可扩展性和伸缩性差等缺点,研究并实现了一种基于Socket分布式计算的泥石流危险性分区系统。引入一个监控节点实时监控当前每个子计算节点的Socket连接数、任务量以及实际处理能力。通过主节点的负载均衡计算实现系统的负载均衡,使得主计算节点更加合理地分配计算任务。实验结果表明,系统完成一次模拟任务所用时间随着并行计算节点数量的增加而减少,具有较高的性价比。

In the past, stand-alone version debris-flow risky district decision system always is stand-alone version. The system has a lot of shortcomings such as slow computing speed, poor scalability and flexibility. Based on socket and distributed computing, a debrisflow risky district decision system is studied and implemented in order to overcome the shortcomings. A monitor computer supervises currently the connection numbers of socket, assignments and the actual processing power of each sub-computing node. A main computingnode calculates the load-information of each sub-computing node to balance the system load. So the main-computing node can distribute computing task more reasonable. Performance test results show that the time of system finishing a simulation task would be less with the increase of number of parallel process nodes and the system has a high price-performance ratio.