頁籤選單縮合
題 名 | On the Efficient Execution of Aggregate Functions for Data Integration in Multidatabases |
---|---|
作 者 | Leu,Yungho; Yang,Chunli; Tang,Shyueming; | 書刊名 | 中華民國資訊學會通訊 |
卷 期 | 3:3 2000.09[民89.09] |
頁 次 | 頁63-78 |
專 輯 | 網際網路與分散式系統 |
分類號 | 312.1 |
關鍵詞 | 資料整合; Data integration; Aggregate functions; Query optimization; |
語 文 | 英文(English) |
英文摘要 | Data integration in a multi-database environment can be done by providing users with an integratedglobal view of the related local databases. An interesting approach for constructing global views isthrough generalization which is modeled as a sequence of outerjoins and aggregations. In this approach,a query on a global view, when posed, is decomposed into a set of subqueries which, in turn, aretransmitted to their corresponding local databases for processing. For this approach to be practical,global queries must be executed efficiently. In this paper, we proposed three useful approaches to optimize the execution of global queries withaggregate functions. The main theme of our approaches is on the localization of the execution of aggregate functions. Through localization, the communication overhead and the processing time required onthe query site can be minimized. In the first approach, we localize the execution of aggregate functionsthrough key value transmission. In the second approach, we maintain maximal values and minimalvalues of all numerical attributes. These values are then used to localize the execution of certain typesof aggregate functions. In the third approach, we maintain a maximal possible difference (MPD) foreach attribute. Similar to the second approach, this value is used to localize the execution of aggregatefunctions. The effectiveness of these approaches is carefully studied and is measured in terms of theratio of data reduced in transmission. |
本系統中英文摘要資訊取自各篇刊載內容。