David J. DeWitt

Professor and Romnes Fellow

Computer Sciences Department
University of Wisconsin
1210 W. Dayton St.
Madison, WI 53706-1685

Telephone: (608) 262-1204
Fax: (608) 262-9777
Email: dewitt@cs.wisc.edu
Ph.D., University of Michigan, 1976
Interests: Object oriented database systems, parallel database systems, database benchmarking, geographic information systems


Research Summary

My two main research projects are SHORE and Paradise. The objective of SHORE is to design, implement, and evaluate a persistent object system that will serve the needs of a wide variety of target applications including hardware and software CAD systems, persistent programming languages, geographic information systems, satellite data repositories, and multimedia applications. SHORE expands on the basic capabilities of the widely-used Exodus Storage Manager (developed at Wisconsin, funded by ARPA) in a number of ways including support for typed objects, multiple programming languages, a `Unix-like' hierarchical name space for named objects, and a Unix-compatible interface to objects with a `text' field. This interface is intended to ease the transition of applications from the Unix file system environment to SHORE as existing Unix tools such as vi and cc will be able to store their data in SHORE objects without modification (basically a Unix file becomes either a single SHORE object or the text field of a complex object). SHORE is being targeted at a wide range of hardware environments, scaling all the way from individual workstations to heterogeneous client/server networks to large multiprocessors such as the Intel Paragon. SHORE is a joint project with Profs. Carey, Naughton, and Solomon.

The Paradise project is attempting to apply the technology developed as part of the SHORE and Gamma projects (Gamma is a parallel relational database system developed at the University of Wisconsin) to the task of storing and manipulating geographic data sets. Currently, many geographic information systems (GIS) use relational database systems to hold their data. While such systems are excellent for managing business data they are a poor match for the modeling needs of a GIS which must be capable of storing and manipulating much more complex objects such as polygons and polylines. Instead, Paradise employs an object-oriented data model, providing a much better match to the type needs of a GIS. Another significant difference from current GIS systems is that Paradise employs parallelism to facilitate executing and processing large data sets such as satellite images. The target hardware platform for the project is a cluster of 64 Sparc 20s connected with ATM.

Sample Recent Publications

The OO7 benchmark (with M. Carey and J. Naughton), Proceedings of the SIGMOD Conference, Washington, DC, May, 1993.

Shoring up persistent applications (with D. DeWitt, M. Franklin, N. Hall, M. McAuliffe, J. Naughton, D. S chuh, C. Tan, O. Tsatalos, S. White, and M. Zwilling), Proceedings of the ACM SIGMOD International Conference on Management of Data, Minneapolis, MN, May, 1994.

Client-server Paradise (with N. Kabra, J. Luo, J. Patel, and J. Yu), Proceedings of the Very Large Data Base Conference, Santiego, Chile, August, 1994.

Recent Talks

VLDB 95 Invited Talk

1996 Object-Relational Summit Presentation


This page was automatically created January 18, 1995.
Email pubs@cs.wisc.edu to report errors.