摘要
随着数据的增长和主流服务器的内存增长,内存数据库作为一种新的数据处理方式受到越来越多的关注.通过研究现有的内存数据库以及其中常见的两种并发控制机制,乐观并发控制(OCC)与悲观并发控制(PCC),分析各自的特点与适应场景.同时提出一种基于已有并发控制的混合并发控制机制(HCC),其中主要包括具体解决两种原有并发控制的兼容问题,并且设计相应的场景信息获取方法以及并发控制的切换方式.最后利用实验比较在静态场景和动态场景下混合并发控制以及两种原有并发控制的性能,具体而言混合并发控制能够在静态场景下接近两种原有并发控制的最优者,而在动态场景下相对于两种原有的并发控制有20%-30%的性能提升.
Due to the data explosion and the memory growth on main-stream servers,the in-memory database is getting more and more attention as a newdata processing method. Based on the study of two common concurrency control mechanisms for in-memory databases,optimistic concurrency control（ OCC） and pessimistic concurrency control（ PCC）,this work finds out their features and suitable scenarios. Meanwhile,this work proposes a hybrid concurrency control（ HCC） on the previous concurrency control mechanisms,which mainly includes the solution to fix the compatibility problem,the mechanism of scenario information collection and the switchover mechanism. Finally this work uses experiments to compare the specific performance of HCC and two previous concurrency control mechanisms in both static and dynamic scenarios,showing that HCC makes suitable choices of the concurrency control mechanisms in static scenarios and achieves 20%-30% performance improvement in dynamic scenarios.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第3期459-463,共5页
Journal of Chinese Computer Systems
关键词
内存数据库
事务并发控制
混合并发控制
并发控制切换
in-memory database
transaction concurrency control
hybrid concurrency control
concurrency control switchover
