基于多路规划遗传算法的负载均衡方法

A Load Balancing Approach Based on Multiprogramming GA

在负载均衡问题中,负载调度方法是核心,它的好坏直接影响均衡系统的性能。提出一种基于多路规划遗传算法的服务器端负载均衡算法。该方法借鉴生物界自然选择和自然遗传机制,模拟自然进化过程搜索最优解,为负载均衡问题提供了新的计算模型。同时,多路规划(多次交叉或变异)后取最优策略的应用,使得多路规划遗传算法的优化性能大为提高。该方法降低了服务器端请求的响应时间,提高了服务器端CPU的利用率,从而改善了系统性能。数据实例表明,该方法是可行的、正确的和有效的。

In the load balancing problem, load schedule algorithm is pivotal. It plays an important role in the whole balancing system. The paper proposes an effective server end load balancing algorithm based on a novel Multi- programming Genetic Algorithm （MGA）. The proposed approach models the natural evolution to search the optimal solution and it is a novel computational model for the load balancing problem. The strategy of multiprogramming （ executes the crossover operation or mutation operation many times ） and choosing the best resultant individuals has largely improved the performance of MGA. Through this schedule algorithm, request response time is lessened, the utility of servers CPU is increased, and performance of the system of load - balancing is enhanced. The final simulation result suggests that this proposed algorithm is feasible, correct and valid.