Distributed Placement of Replicas in Hierarchical Data Grids with User and System QoS Constraints
Distributed Placement of Replicas in Hierarchical Data Grids with User and System QoS Constraints
复制标题
DOI：
10.1109/3pgcic.2011.35
复制
发表时间：
2011-10
期刊：
影响因子：
--
通讯作者：
Mohammad Shorfuzzaman;P. Graham;Mehmet Rasit Eskicioglu
中科院分区：
文献类型：
--
作者：
Mohammad Shorfuzzaman;P. Graham;Mehmet Rasit Eskicioglu
数据网格支持需要访问存储在世界各地的海量数据集的分布式数据密集型应用程序。广域网的高延迟阻碍了确保有效访问此类数据集。为了加快访问速度，可以复制文件，以便用户可以访问附近的副本。关于数据网格中副本放置问题的大部分工作都集中在平均系统性能上，而忽略了质量保证问题。在考虑QoS的现有工作中，通常假设简化的复制模型，因此，所得的解决方案可能不适用于实际系统。在本文中，我们引入了一种更现实的分层数据网格中副本放置模型，该模型从用户和系统的角度确定了预期满足某些质量要求的最小副本数量的位置。我们的放置算法基于高度分布式和去中心化的技术，该技术利用流行数据文件的数据访问历史记录，并通过最小化总体复制成本（读取和更新）来计算副本位置，同时最大化给定流量模式的 QoS 满意度。该问题是使用动态规划来表述的。我们使用 OptorSim 评估我们的算法。仿真结果证明了我们的副本放置技术的有效性，考虑了副本服务器的存储和工作负载限制、链路容量限制、用户 QoS 要求等各种因素。
Data grids support distributed data-intensive applications that need to access massive datasets stored around the world. Ensuring efficient access to such datasets is hindered by the high latencies of wide-area networks. To speed up access, files can be replicated so a user can access a nearby replica. Much of the work on the replica placement problem in data grids has focused on average system performance and ignored quality assurance issues. In the existing work that considers QoS, a simplified replication model is often assumed, therefore, resulting solutions may not be applicable to real systems. In this paper, we introduce a more realistic model for replica placement in hierarchical Data Grids which determines the positions of a minimum number of replicas expected to satisfy certain quality requirements both from user and system perspectives. Our placement algorithm is based on a highly distributed and decentralized technique that exploits the data access history for popular data files and computes replica locations by minimizing overall replication cost (read and update) while maximizing QoS satisfaction for a given traffic pattern. The problem is formulated using dynamic programming. We assess our algorithm using OptorSim. Simulation results demonstrate the effectiveness of our replica placement technique considering various factors such as storage and workload constraints of replica servers, link capacity constraints, user QoS requirements, etc.