云环境下基于MapReduce的海量服务选择研究

RESEARCH ON THE MASSIVE SERVICES SELECTION BASED ON MAPREDUCE IN CLOUD ENVIRONMENT

当处理分布式、大规模的服务选择时,传统服务选择方法存在着效率不高和全局Qo S性能低下的问题。基于Map Reduce框架,设计了一种云环境下的海量服务选择方法以解决此问题。首先,基于Map Reduce框架,利用Skyline算法,筛选海量候选服务,生成Skyline服务库;其次,基于迭代式Map Reduce框架,运用多目标模拟退火算法,从所生成的Skyline服务库中优选Skyline服务,产生一组Pareto最优的组合服务;最后,依据用户的个性化和多样性需求,执行Top-k查询,优选出满足用户偏好的k个组合服务。该方法适应于具有分布式环境、高维Qo S的海量服务选择,能快速返回组合服务,且其全局Qo S较优。

When dealing with distributed and massive services selection, traditional approaches of service selection have lower efficiency and poorer performance of global QoS. We present an approach of massive services selection based on MapReduce framework in cloud environment to solve the problem. Firstly, we screen massive candidate services and generate a library of Skyline services by using Skyline algorithm based on the MapReduce framework; Secondly, we select the preferred Skyline services form the generated Skyline services library to generate a set of Pareto optimal composite services using multi-objective simulated annealing algorithm based on iterative MapReduce framework; Finally, according to the user＇s personalized and diverse demand, we execute Top-k queries to select preferablyk composite services which meet the user＇s preference. The proposed approach is adapted to service selection with services of large-scale, high-dimensional QoS in distributed environment, it can quickly return to composite services, and its global QoS are optimum.