Task scheduling mechanism is one of the core issues in cloud computing. The task scheduling algorithm in cloud computing re-quires improvement of the system throughput and the largest span while considering resources security and load balancing problems. As a classical task scheduling algorithm with powerful and implicit parallel space search capability,genetic algorithm is widely used in cloud computing. However,it has many deficiencies,such as slow convergence and premature with the increasing calculation scale. Min-Min al-gorithm and Max-Min algorithm are simple and practicable with better makespan,which can well make up the deficiencies of traditional genetic algorithm. On this basis,an improved algorithm is put forward,which introduces Min-Min algorithm and Max-Min algorithm in the process of population initialization,and uses the minimizing makespan and the load balancing of resource as double-fitness function meanwhile. The simulation shows that this algorithm can elevate the quality of initial population,the search capability and the convergence rate,which is more efficient.%任务调度是云计算的核心问题。云计算中的任务调度算法要求在提高系统吞吐量和最大跨度的同时又要兼顾资源的安全与负载均衡问题。传统遗传算法因具有强大的并行空间搜索能力而在云计算中得到广泛应用，但其亦存在明显不足，即随着计算机规模的不断扩大，收敛性逐渐降低，存在易早熟等不足，限制了其调度性能。而Min-Min和Max-Min算法简单易行，且具有较好的时间跨度，可以较好地弥补传统算法的不足。在传统遗传算法的基础上，结合Min-Min和Max-Min算法，提出了一种新的云计算任务调度算法，在产生初始化种群时引入Min-Min和Max-Min算法，并选取任务完成时间和负载均衡作为双适应度函数，提高了初始化种群的质量、算法搜索能力以及收敛速度。仿真结果表明，该算法优于传统遗传算法，是一种有效的云计算任务调度算法。
展开▼
