计算机应用研究2024，Vol.41Issue(1)：226-230,253,6.DOI:10.19734/j.issn.1001-3695.2023.04.0188
面向持久化键值数据库的自适应热点感知哈希索引
Adaptive hot-aware hash index for persistent key-value databases
摘要
Abstract
In view of the lack of hotspot awareness in existing key-value database storage systems,which leads to poor perfor-mance and unreliability under highly tilted workloads,this paper proposed an adaptive hot-spot aware hash index model,which implemented a high-performance hash table based on key summary information.Firstly,the paper used the abstract information of key to replace the key value,compressed the storage space of key,and optimized the data structure of the bucket in the hash table.Secondly,the paper optimized the probe operation of hash table by using the data level parallel technique of CPU and CPU cache line.Finally,in order to solve the problem of extra disk I/O due to the inaccurate comparison of key values caused by summary information,this paper designed an adaptive key scheduling algorithm,which dynamically adjusted the storage lo-cation of key values according to the current available memory size,hash index load and access hotspot.Experiments on YCSB simulation datasets show that the adaptive hot-spot aware hash index is up to 1.2 times faster than the most advanced hash ta-ble for the same memory usage.关键词
持久化键值存储/自适应/热点感知/哈希索引Key words
persistent key-value store/self-adaptation/hot spot perception/hash index分类
信息技术与安全科学引用本文复制引用
王楠,吴云..面向持久化键值数据库的自适应热点感知哈希索引[J].计算机应用研究,2024,41(1):226-230,253,6.基金项目
贵州省科学技术厅资助项目(黔科合基础-ZK[2022]—般119) （黔科合基础-ZK[2022]—般119）
