Abstract: Replica technology is widely applied in cloud computing and distributed systems, the reasonable data replica placement is an important means to reduce the operation cost of network, and is also the key issue of replica technology. Replica update is that the replicas are added and deleted when data access requirements dynamically change. Based on multiple policy, this paper proposes a dynamic updating algorithm named MPFSF (min_placement far servers first) to solve the update problem. According to the need to implement the necessary replica update, this paper reuses as much as possible the pre-existing replicas and maximizes the capacity of each replica to reduce the date access latency and improve the efficiency of data access, by adding the constraint of communication distance. The experimental results show that, for the proposed algorithm with update policy, the running cost of the network gets a considerable reduction under the different distance limits, and the best improvement is up to 84.6% in comparison to the state-of-the-art.

Key words: tree network, replica placement, multiple policy, update

摘要： 副本技术广泛应用于云计算及分布式系统中，合理的数据副本放置是降低网络运行成本的重要手段，也是副本技术的核心问题。副本更新是针对网络中数据访问请求的动态变化而进行的副本添加与删除。针对副本放置问题，提出了一种基于多访问策略的副本动态更新算法MPFSF（min_placement far servers first）。该算法在引入通信距离限制的同时，尽可能多地重用网络中已存在的副本，并根据需要实施必要的副本更新，最大使用每个副本的处理能力，以便降低数据访问的时延，提高数据访问效率。最后通过实验结果和分析证明，该算法在不同的通信距离限制下，网络的运行成本得到了明显的降低，对原有算法的改进最高可达84.6%。

关键词: 树形网络, 副本放置, 多访问策略, 更新