云数据中心操作系统副本分布算法的设计与实现

摘要

The thesis introduces a replica distrubution algorithm in Yunhai OS, a cloud data center operation system. The algorithm is used to solve the replica distribution problem under cloud storage environments by transforming storage node selection problem into multi-criteria decision making problem (MCDM) which is hence solved by TOPSIS. The algorithm can take full advantage of various data monitored under cloud computing environments, integrate flexible weight allocation methods, and suits most cloud storage environments. Simulation experiments illustrate that Yunhai OS algorithm is superior to traditional Least and Random algorithms at both load balance and replica creation time.%介绍云数据中心操作系统(云海OS)中的副本分布算法,该算法用于解决云存储环境下的副本分布问题,将存储节点的选择问题转化为一个多指标决策问题(MCDM),使用TOPSIS进行求解.算法能够充分利用云计算环境下的多种检测数据,结合灵活的权重分配方式,适应多数云存储环境.模拟实验表明,云海OS算法在负载均衡和副本创建时间方面优于传统的Least和Ran-dom算法.