基于改进多时隙散点算法的大数据任务调度

Big Data Scheduling Management Method Based on Improved Multi-time-slot Spread Nodes Algorithm

针对当前大数据任务的调度管理,传统的调度处理方法采用随机调度机制,从单个节点的性能出发,无法达到全局最优化的效果。提出基于改进多时隙散点算法的大数据任务调度管理方法,为了克服多时隙调度算法的缺点,运用多时隙调度算法时,引入了一个管理因子,对系统节点进行数据调度任务分配,根据分配的任务进行评价和估计,并且不断更新实时任务,追求最优性能,最终达到全局最优化。最后通过一组20节点数据进行测试实验,结果显示,采用改进的多时隙大数据任务调度管理算法,相对于传统调度方法,系统整体效率提高了平均约28%,具有很好的应用价值。

The traditional scheduling methods using random scheduling mechanism, starting from the performance of a sin-gle node, cannot reach the effect of the global optimization;Is proposed based on improved scatter algorithm of contention slot more big data scheduling management methods, first of all, in order to get rid of disadvantages of multiple time slot scheduling algorithm, using more time slot scheduling algorithms, introduces a management factor, the nodes of the system for data scheduling task allocation, and then evaluate the assigned tasks and estimates, and constantly updated real-time tasks, the pursuit of optimal performance, finally achieve global optimization. Finally through a set of 20 node data testing experiment, the results show that, with the improved much time slot scheduling management, compared with traditional scheduling method, the system overall efficiency is increased by an average of about 28%, it has the very good application value.