可平行化之資料倉儲索引技術

Abstract

資料倉儲的概念主要是由數個獨立的資料庫中，收集有效資訊，先進行初步整合後，放置於資料中心，並提供企業決策支援分析作業之使用。資料倉儲及線上分析作業已是決策系統中不可或缺的組成元件，而在資料倉儲大量資料上所做的查詢可能是很複雜以致查詢相當費時，因此如何在資料倉儲上建製可大大降低查詢時間的索引技巧，降低查詢時間達到即時回傳決策者所需的建議答案是相當重要的。 本篇論文主在提出可平行化的索引技術，針對資料倉儲及線上分析作業不同的模組需求提出三個可提供快速線上分析作業且平行化的索引技術，分別為簡單的bit-wise索引技術、可壓縮的bit-wise索引技術、以查詢頻率為基準的bit-wise索引技術。最後，提出相對應的平行演算法。

Data warehouse is an information provider that collects necessary data from individual source databases to support the analytical processing of decision-support functions. Data warehousing (DW) and online analytical processing (OLAP) are becoming critical components in decision support system. The queries of OLAP/OLTP requirements in a data warehouse may be very complex and time consuming, a good indexing strategy that can reduce the query time is necessary for the users of the data warehouses. In this thesis, the new, efficient and parallelized indexing methods based upon bit-wised indexing for data warehousing will be proposed to reduce the overhead. Three indexing models, including simple bit-wise indexing method, condensable bit-wise indexing method and frequency-based bit-wised indexing method, are proposed for three types of data warehousing and OLAP environments requires. Also, the corresponding indexing and matching algorithms for such indexing models are also proposed. Finally, the parallelized issues of three indexing methods are proposed.