DocumentCode
2445138
Title
XPCache: An Efficient Query Processor for Client-Server Architecture
Author
Taha, Kamal ; Elmasri, Ramez
Author_Institution
Dept. of Comput. Sci. & Eng., Univ. of Texas at Arlington, Arlington, TX
fYear
2007
fDate
15-15 April 2007
Firstpage
49
Lastpage
54
Abstract
This study addresses the problem of distributed XML queries in client-server architecture. Distributed XML documents can be large and complex. Due to the overhead involved in decomposing the queries, sending the decomposed queries to remote site(s), and executing structural join operations to compose the results, it can be difficult to rapidly execute queries on these distributed XML docs. We propose a query processor called XPCache to tackle those problems. We describe a Global Query Graph, a novel mechanism used by XPCache to classify queries based on their type and frequency. We then describe XPCache´s mechanisms that exploit those classifications to improve query response time. Queries classified under the same type are given the same query optimization plan. These plans are cached at the central site. Queries classified as frequently used are answered from the central site cache, which improves query response time. The performance of XPCache is compared with another system and the results showing marked performance improvement.
Keywords
XML; cache storage; client-server systems; pattern classification; query processing; XPCache; client-server architecture; distributed XML query processing; global query graph; pattern classification; query optimization plan; query response time; structural join operation; Computer architecture; Computer science; Delay effects; Frequency; Middleware; Pattern matching; Query processing; Tree graphs; XML;
fLanguage
English
Publisher
ieee
Conference_Titel
Databases for Next Generation Researchers, 2007. SWOD 2007. IEEE International Workshop on
Conference_Location
Istanbul
Print_ISBN
1-4244-0903-9
Electronic_ISBN
1-4244-0904-7
Type
conf
DOI
10.1109/SWOD.2007.353197
Filename
4163061
Link To Document