论文部分内容阅读
有效的数据处理是大规模分布式环境(如云数据)中的一个关键性问题,其中需要考虑到数据的复制。数据复制可以减少服务时间和获取数据所需的时间,增加可用性并优化系统负载分布。然而值得一提的是,数据的同样会增加储存数据所需的资源和能源。我们提出了一种可减少资源、能源消耗,减少系统延迟,并增加系统可用性的复制管理器,称为位置复制管理器(Locality replication manager,LRM)。这一管理器采用的两种重要算法利用了数据块之间的物理邻接特性。对LRM进行的一系列模拟结果显示,LRM消耗了较少的资源和能源,优化了系统负载分布,并增加了系统可用性,减少了系统延迟,因此适用对于分布式系统。
Effective data processing is a key issue in large-scale distributed environments such as cloud data, where data replication needs to be considered. Data replication can reduce service time and data acquisition time, increase availability and optimize system load distribution. However, it is worth mentioning that the data will also increase the resources and energy needed to store the data. We propose a replication manager that reduces resources, energy consumption, reduces system latency, and increases system availability, called the Locality replication manager (LRM). Two important algorithms used by this manager take advantage of the physical proximity between the data blocks. A series of simulations of the LRM show that the LRM consumes less resources and energy, optimizes system load distribution, increases system availability, and reduces system latency, making it suitable for distributed systems.