大规模混合数据库缓冲区替换方法优化仿真  

Substitution Method of Large Scale Hybrid Database Buffer Optimization Simulation

作　　者：陆云帆[1] 诸振家[1] 孙洁[1] 王鑫[2] 

机构地区：[1]深圳信息职业技术学院软件学院,广东深圳518000 [2]深圳大学机电与控制工程学院,广东深圳518060

出　　处：《计算机仿真》2016年第8期425-429,共5页Computer Simulation

基　　金：校级科研项目"基于云计算的债券收益率大数据分析系统研究"(lg201429)

摘　　要：为提高数据库数据使用率,对大规模混合数据库的缓冲区进行替换优化。由于数据库存储数据量较大,使得需要多个缓冲区进行数据间的替换。传统的替换方法,主要在数据传输过程中对缓冲区直接进行替换,忽略了数据滞留等原因造成的缓冲区替换不全面的问题。提出生命值敏感的大规模混合数据库缓冲区替换优化方法。对数据库缓冲区进行划分,定义数据库缓冲区内各个数据页的访问频度、新颖度和读写代价权重的生命值,将频率因素引入传统替换算法中,依据用户实时访问请求和历史访问记录预测其将来可能要访问的数据,并形成用户访问数据页预测对象集,利用改进的传统替换算法将权值较弱且不属于预测对象集的数据页进行替换。实验结果表明,采用改进方法相比传统替换方法,可有效提高数据库使用率。It can improve data utilization rate in large scale hybrid database to optimize buffer replacement. The storage data volume is huge in database. It needs lots of buffers to replace data. Traditional method replaces buffer directly during data transmission. It ignores the problem of incomplete buffer replacement caused by data retention.In this paper,we propose an optimization replacement method of buffer in large scale hybrid database. It is sensitive to life value. Firstly,we divided buffer in data base and define accessing frequency,novelty and weight life value at read-write cost of each data page in database buffer. The frequency factor is introduced into traditional replacement algorithm. Then we predict the data to be accessed probably according to real-time access request and history access record. We also build prediction object set of user data access page. Finally,we use the modified algorithm replaced data page of smaller weights and not belonging to prediction object set. The simulation results show that the modified method can improve utilization rate of database compared with traditional method.

关 键 词：混合大规模 数据库 缓冲区替换 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]