Colloquium Speaker

Speaker: Fatma Ozcan
University of Maryland
Topic:Improving the Performance of Heterogeneous Databases
Date:Tuesday, March 20, 2001
Time:11:00 AM
Place:Gould-Simpson, Room 701

Refreshments will be served in the 7th-floor lobby of Gould-Simpson at 10:45 AM


My work focuses on optimizing the performance of servers that access multiple networked heterogeneous data sources and that process a large number of concurrent submitted queries. In particular, I have developed techniques that address the following three problems: (1) Given a networked query that accesses heterogeneous data sources, how do we compute its cost? (2) Given a set of queries spanning heterogeneous networked data sources, how do we merge these queries together so as to reduce the workload on the server? and (3) Given a set of queries, how can we partition this set of queries into clusters so that the queries in each cluster may be merged using the techniques of (2) above? Specifically, I will summarize our solutions to problems (1) and (2) and describe algorithms which partition a large number of concurrent queries into components of small size (e.g. 10 to 50) so that each component's queries can be merged using techniques we developed to solve problem (2). I will also provide experimental results that justify the scalability of our approach.