Development of new data partitioning and allocation algorithms for query optimization of distributed data warehouse systems

Distributed databases and in particular distributed data warehousing are becoming an increasingly important technology for information integration and data analysis. Data Warehouse (DW) systems are used by decision makers for performance measurement and decision support. However, although data warehousing and on-line analytical processing (OLAP) are essential elements of decision support, the OLAP query response time is strongly affected by the volume of data need to be accessed from storage disks. Data partitioning is one of the physical design techniques that may be used to optimize query processing cost in DWs. It is a non redundant optimization technique because it does not replicate data, contrary to redundant techniques like materialized views and indexes. The warehouse partitioning problem is concerned with determining the set of dimension tables to be partitioned and using them to generate the fact table fragments. In this work an enhanced grouping algorithm that avoids the limitations of some existing vertical partitioning algorithms is proposed. Furthermore, a static partitioning algorithm that allows fragmentation at early stages of schema design is presented. The thesis also, investigates the performance of the data warehouse after implementing a combination of Genetic Algorithm (GA) and Simulated Annealing (SA) techniques to horizontally partition the data warehouse star schema. It, then presents the experimentation and implementation results of the proposed algorithm. This research presented different approaches to optimize data fragments allocation cost using a greedy mathematical model and a combination of simulated annealing and genetic algorithm to determine the site by site allocation leading to optimal solutions for fragments distribution. Throughout this thesis, the term fragmentation and partitioning will be used interchangeably.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

London Metropolitan University

Year dc:date.issued

2008

Author dc:creator

• Abdalla, Hassan Ismail

• 000 Computer science, information & general works

Dc Identifier Grantnumber

N/A

OAI identifier oai:identifier

oai:repository.londonmet.ac.uk:8333