文章基本信息

标题：A Rough-Columnar RDBMS Engine ¨C A Case Study of Correlated Subqueries
本地全文：下载
作者：Dominik ¡ä Sle.zak University of Warsaw ; Infobright Inc. ; Poland [email protected] Piotr Synak Infobright Inc. 等
期刊名称：Bulletin of the Technical Committee on Data Engineering
出版年度：2012
卷号：35
期号：01
出版社：IEEE Computer Society
摘要：Columnar databases provide a number of benefits with regard to both data storage (e.g.: data compres- sion) and data processing (e.g.: optimized data access, parallelized decompression, lazy materialization of intermediate results). Their characteristics are particularly advantageous for exploratory sessions and ad hoc analytics. The principles of columnar stores can be also combined with a pipelined and iterative processing, leading toward modern analytic engines able to handle large, rapidly growing data sets. In this paper, we show how to further enrich such a framework by employing metadata layers aimed at minimizing the need of data access. In particular, we discuss the current implementation and the future roadmap for correlated subqueries in Infobright¡¯s RDBMS, where all above-mentioned archi- tectural features interact with each other in order to improve the query execution.