摘要： Key-Value(KV)是NoSQL系统中使用较为广泛的一种存储模型。针对当前主流NoSQL系统存在检索功能有限、内存容量小和宕机等问题,提出一种KV型NoSQL系统设计方法。使用可持久化的混合主索引结构解决范围查询和快速重启,采用数据分布算法和内外存混合存储技术,通过数据在内存和固态硬盘之间的智能调度,实现了大规模数据的高效存取和低成本存储。实验结果表明,与Redis相比,对于典型大小Value,该系统在单节点读写性能上提升约8倍,配置成本降低约3/4,在海量数据的低成本高效存储上有明显优势。

关键词: NoSQL系统, 多核架构, 固态硬盘, 持久化存储, 数据分布, 混合存储

Abstract: The Key-Value(KV) is a storage model widely used in NoSQL systems.To address the problems of limited search function,small memory capacity and downtime of the current mainstream NoSQL system,this paper proposes a KV type of NoSQL system design method.The method designs a persistent hybrid primary index structure to solve range queries and fast restarting,uses data distribution algorithms and internal and external hybrid storage technologies,and implements efficient access and low-cost storage of large-scale data through intelligent scheduling of data between memory and Solid State Disk(SSD).Experimental results show that compared withRedis,the system improves the single-node read/write performance by about 8 times and reduces the configuration cost by about three-fourths when addressing values of typical size.It has obvious advantages in the low-cost and efficient storage of massive data.

Key words: NoSQL system, multi-core architecture, Solid State Disk(SSD), persistent storage, data distribution, hybrid storage

中图分类号: 

• TP391