Indexing and Reorganization in Parallel Database Systems

Northeastern University, Boston MA

Investigators

Abstract

Indexing and Reorganization in Parallel Database Systems The goal of this research project is to design, analyze and simulate restartable algorithms for on-line reorganization in parallel database management systems (DBMSs). The growing need for 7/24 (7 days a week/24 hours a day) database access for web applications makes development of these algorithms urgent. Applications such as those found in telecommunications and web-based businesses (for example, eBay or AOL) require very large highly available databases. These are or will soon be managed by parallel DBMSs (i.e., DBMSs on clusters of workstations, each managing its own collections of disks, connected by Local Area Networks (LANs)), so parallel algorithms to maintain availability in the face of logical and physical changes to the database as well as unexpected growth are important. The approach deployed in the project includes the description of explicit logging operations to assure consistency and preserve incremental improvements in the face of possible system failures at any node of the parallel system. Collaboration with industry is part of the project. Ms. Xiaowei Sun, a supported Ph.D. student on the project, has completed an internship at Clustra, Inc., whose product is a parallel replicated DBMS designed for high availability. She is also doing an internship with IBM Toronto on online reorganization for IBM's DBMS, DB2. The close contact with industry and the current importance of the problem of online reorganization will strengthen the impact of this project's results, including the expected technology transfer. http://www.ccs.neu.edu/home/salzberg/nsfproj2000.html