| 引用本文: | 刘仲,任浩,周兴铭.基于镜像的高可用数据对象布局算法.[J].国防科技大学学报,2006,28(5):42-46.[点击复制] |
| LIU Zhong,REN Hao,ZHOU Xingming.High Availability Data Objects Placement Algorithm with Mirroring[J].Journal of National University of Defense Technology,2006,28(5):42-46[点击复制] |
|
| |
|
|
| 本文已被：浏览 7197次 下载 5859次 |
| 基于镜像的高可用数据对象布局算法 |
| 刘仲, 任浩, 周兴铭 |
|
(国防科技大学 计算机学院，湖南 长沙 410073)
|
| 摘要: |
| 镜像是提高数据可用性的流行方法。借鉴RAID的方法，在算法一级实现数据的冗余分布，提出基于镜像的高可用数据对象布局算法。在数据对象和存储节点失效时，利用冗余数据重构数据对象和存储节点，有效保证存储系统的高可用性。采用马尔可夫激励模型对存储系统进行定量的可用性分析，计算结果表明该方法是有效的。 |
| 关键词: 大规模存储系统 高可用性 数据对象布局 |
| DOI： |
| 投稿日期：2006-06-01 |
| 基金项目:国家自然科学基金资助项目(60503042) |
|
| High Availability Data Objects Placement Algorithm with Mirroring |
| LIU Zhong, REN Hao, ZHOU Xingming |
|
(College of Computer, National Univ. of Defense Technology, Changsha 410073, China)
|
| Abstract: |
| Mirroring is a popular technique for enhancing data availability. Based on this technique, this paper proposes a high-availability data objects placement algorithm with mirroring, which groups objects into redundancy sets by using RAID at the algorithm level. The redundancy allows us to reconstruct any corruption or failure of the data objects and storage nodes, thus efficiently ensuring the high availability of storage system. We quantify the availability of storage system by using Markov reward model, and the computing results indicate that the algorithm is efficient. |
| Keywords: large-scale storage system high availability data objects placement |
|
|
|
|
|
