计算机科学
R-树
树(集合论)
节点(物理)
B-树
索引(排版)
算法
二进制对数
随机存取
理论计算机科学
空间数据库
空间分析
离散数学
二叉树
组合数学
数学
统计
操作系统
结构工程
工程类
万维网
作者
Yanfei Lv,Jing Li,Bin Cui,Xuexuan Chen
标识
DOI:10.1007/978-3-642-20244-5_20
摘要
R-Tree structure is widely adopted as a general spatial index with the assumption that the deployed system is equipped with magnetic hard disk. While the application of SSD becomes more and more popular, traditional optimization of R-Tree structure on SSD is much less desirable than that on magnetic hard disk. Existing flash-aware index approaches employ log mechanism to reduce random writes at a cost of large amount of read. A novel index named Log Compact R-Tree (LCR-tree) is proposed in this paper. Distinguished from previous attempts, compacted log is introduced to combine newly arrival log with origin ones on the same node, which renders great decrement of random writes with at most one additional read for each node access. Extensive experiments illustrate that the proposed LCR-Tree can achieve up to 3 times benefit against existing approaches.
科研通智能强力驱动
Strongly Powered by AbleSci AI