home · mobile · calendar · colloquia · 1999-2000 · 

Colloquium - Osborne

The Sanctuary Data Integration Environment
Richard Osborne
Department of Computer Science, University of Colorado
4/27/2000
3:30pm-4:30pm

Modern persistent applications typically run on top of several (if not hundreds) of distributed, heterogeneous databases (e.g., legacy databases and modern relational and object-oriented databases, as well as ad hoc systems). Integrating, evolving and maintaining such a set of databases is a huge problem which has traditionally been attacked in one of two ways: point-to-point solutions and complete integration. Point-to-point solutions consist of building a new set of wrappers or applications each time a new database needs to be accessed. Maintaining data consistency between a large set of heterogeneous databases in such an ad hoc manner is generally doomed to failure. Fully integrated systems solve many of the consistency and integrity issues associated with point solutions due to the fact that the database schemas are typically translated into a common model so that they may be merged into a common view. Unfortunately, for an application involving a large number of rapidly evolving databases (with new ones being frequently added), this solution is generally intractable.

Sanctuary takes a different approach by shifting the main workload of the integration process from the client to Sanctuary itself, and performing that integration in a "lightweight" manner. Sanctuary (formerly Sybil) is a data source integration environment that works on top of distributed object technologies. Sanctuary was successfully used in several integration processes of different heterogeneous data sources. With Sanctuary, integration is shifted from the client to special objects, called mediators, and is performed using a language and environment specialized to the task of performing data integration. Instead of dealing with several interfaces and performing the integration work on the client side, a Sanctuary mediator provides the user with a single (user-defined) interface and the integration of the data from the heterogeneous data sources is handled by the system.

Refreshments will be served in ECOT 831 immediately following the talk.
Hosted by Roger (Buzz) King.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
webmaster@cs.colorado.edu
www.cs.colorado.edu
May 5, 2012 (14:13)
XHTML 1.0/CSS2
©2012