计算机科学 ›› 2014, Vol. 41 ›› Issue (6): 142-147.doi: 10.11896/j.issn.1002-137X.2014.06.028
丁祥武,张光辉
DING Xiang-wu and ZHANG Guang-hui
摘要： 片上多核处理器(CMP)凭借其低功耗和低成本等优势迅速成为处理器市场的主角,它为多线程的实现提供了硬件支持。列存储技术在分析型应用中具有显著的优势。在列存储系统中,查询优化依然是最重要的问题之一。在列存储系统中,利用多核资源提高查询处理性能具有较大潜力。文中通过对查询执行器生成的物理查询树进行流水多线程设计,结合列存储的特点,建立传递块缓冲区,使主线程和辅助线程分别对传递块缓冲区读写,以提高查询性能。同时还提出使用操作系统中经典的“生产者和消费者”模式来解决线程之间的同步问题。提出的这些方法应用在实验室研发的列存储系统DWMS中,使用数据仓库基准测试集SSB验证了这些方法的有效性。实验结果表明,传递块缓冲区的设计使SQL的查询效率有了近50%的提升。
| [1] Copeland G P,Khoshafian S N．A decomposition storage model[C]∥Proceedings of SIGMOD International Conference．Austin,Taxes:ACM,1985:268-279 [2] MacNicol R,French B．Sybase IQ multiplex-Designed for analy-tics[C]∥Proceedings of the 30th Very Large Data Base Confe-rence．Toronto,Canada:VLDB Endowment,2004:1227-1230 [3] Stonebraker M,Abadi D J,Batkin A,et al．C-Store:A column-oriented DBMS[C]∥Proceedings of the 31st Very Large Data Base Conference．Trondheim,Norway:VLDB Endowment,2005:553-564 [4] Boncz P A．Monet:A next-generation DBMS kernel for query-intensive applications[D]．Amsterdam:Universiteit van Amsterdam,2002 [5] Hennessy J L,Patterson D A．Computer Architecture(4th ed)[M].Morgan Kaufman Publishers,2007 [6] Abadi D J.Query Execution in Column-Oriented Database System[D]．MIT PhD Dissertation,2008 [7] Blanas S,Li Yi-nan,Patel J M.Design and Evaluation of Main Memory Hash Join Algorithms for Multi-core CPUs[C]∥Proceedings of the ACM SIGMOD Conference.Athens,Greece:ACM,2011 [8] Zhou Jing-ren,Cieslewicz J,Ross K A,et al.Improving database performance on simultaneous multithreading processors[C]∥Proceedings of the 31st international conference on Very large data bases.Trondheim,Norway:VLDB Endowment,2005:49-60 [9] Cieslewicz J,Ross K A,Giannakakis I.Parallel buffers for chip multiprocessors[C]∥Proceedings of the 3rd International Workshop on Data Management on new Hardware．New York,USA:ACM,2007 [10] Garcia P,Madison,WiHenry,et al.Pipelined hash-join on multithreaded architectures[C]∥Proceedings of the 3rd internationalworkshop on Data management on new hardware.NewYork,USA:ACM,2007 [11] Nehme R,Bruno N.Automated Partitioning Design in Parallel Database Systems[C]∥ Proceedings of the ACM SIGMOD Conference.Athens,Greece:ACM,2011:1137-1148 [12] 汤子瀛,哲凤屏,汤小丹.计算机操作系统[M].西安:西安电子科技大学出版社,2002 [13] Harizopoulos S,Shkapenyuk V,Anastassia.QPipe:a simultaneously pipelined relational query engine[C]∥Proceedings of the 2005ACM SIGMOD international conference on Management of data.New York,USA:ACM,2005:383-394 [14] Hardavellas N,Pandis I,Johnson R,et al.Database servers onchip multiprocessors:Limitations and opportunities[C]∥CIDR．2007:79-87 [15] Manegold S,Boncz P,Nes N,et al.Cache-conscious radix-de-cluster projections[C]∥Proceedings of the Thirtieth International Conference on Very Large Data Bases.VLDB Endowment,2004:684-695 [16] Rao Jun,Zhang Chun,Megiddo N,et al.Automating physical database design in a parallel database [C]∥Proceedings of the 2002ACM SIGMOD Interna-tional Conference on Management of data.New York,USA:ACM,2002:558-569 [17] Cieslewicz J,Mee W,Ross K A．Cache-conscious buffering for database operators with state [C]∥Proceedings of the Fifth International Workshop on Data Management on New Hardware.New York,USA:ACM,2009:43-51 [18] 吴峻峰,许跃生,张永东,等.CC＄:一种面向分布式众核平台的并行编程语言[J].计算机科学,2013,40(3):128-132 [19] O’Neil P,O’Neil B,Chen Xue-dong．Star schema bench-mark Revision 3June 5[EB/OL]．http://www.cs.umb.edu/~poneil/,2010-02 |
| No related articles found! |
|
||
首页