摘要
随着社会信息化程度的不断提高,各种形式的数据急剧膨胀。HDFS成为解决海量数据存储问题的一个分布式文件系统,而副本技术是云存储系统的关键。提出了一种基于初始信息素筛选的蚁群优化算法(Init Ph_ACO)的副本选择策略,通过将遗传算法(GA)与蚁群优化算法(ACO)算法相结合,将它们进行动态衔接。提出基于初始信息素筛选的ACO算法,既克服了ACO算法初始搜索速度慢,又充分利用GA的快速随机全局搜索能力。利用云计算仿真工具Cloud Sim来验证此策略的效果,结果表明:Init Ph_ACO策略在作业执行时间、副本读取响应时间和副本负载均衡性三个方面的性能均优于基于ACO算法的副本选择策略和基于GA的副本选择策略。
With the degree of social information continues to improve, various forms of data expand rapidly. Hadoop distributed file system （HDFS） has become a distributed file system solving mass data storage problem, and a copy of the technical is the key of cloud storage system. Present a copy selection strategy foundation on ant colony optimization algorithm based on initial pheromone screening （InitPh_ACO）strategy, by combining genetic algorithm （GA） and ant colony algorithm, link them dynamically and propose ant colony algorithm based on initial screening of pheromone. This algorithm not only overcome shortage of slow initial search of ant colony algorithm, and make full use of fast stochastic global search capability of GA. Using cloud computing simulation tools CloudSim to verify the effect of this strategy, the results show that InitPh_ACO strategy are prior to the selection strategy replica ACO strategy algorithm and a copy of the selection strategy based on GA policy in three aspects of performance which are job execution time, response time and the copy of load balancing.
作者
段效琛
李英娜
贾会玲
赵振刚
李川
DUAN Xiao-chen LI Ying-na JIA Hui-ling ZHAO Zhen-gang LI Chuan(Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming 650500, Chin)
出处
《传感器与微系统》
CSCD
2017年第4期31-33,38,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(51567013)
