Active Databases are the proposed solution for applications that require timely response to spontaneously arising, critical situations. Typical examples are network management, CIM, commodity trading, air-traffic control, and various tracking applications. Conventional database management systems deal with such situations by frequent polling of he database, or by embedding the situation monitoring capabilities in the application, leading to loss of modularity and poor application maintainability. The way current DBMSs deal with these applications is comparable to implementing one's own data structures and file handling instead of storing data in a database. The improvement in responsiveness and software quality through the use of active databases is comparable to the quantum leap achieved years ago through the use of databases instead of file systems and ad-hoc data structures.

Active databases are centered on the embedding of event-condition-action rules in the database system. This allows the homogeneous definition of rules and their consistent administration. It further sets the basics for expressing and implementing a variety of DBMS components, such as the access control systems, the transaction manager, the integrity manager, and the view manager in a homogeneous and extensible manner.

The time-critical nature of applications that have been considered as prime candidates to benefit from active database technology also implies the importance of real-time aspects and timing constraints in active databases. The proper integration of time handle temporal events and timing constraints is crucial.

While progress has been achieved over the past 4 years, most proposed solutions deal only with small subproblems. In addition to the important integration issues, many open problems remain. These range from definition of declarative, user-understandable rule representations, to consistency verification among sets of rules, to efficient event monitoring and transaction mechanisms that can correctly execute triggered actions. Topics addressed include but are not limited to:

• Architecture
• Event specification
• Optimization
• Execution Models/Transaction Models
• Representation of timing constraints
• Execution under timing constraints
• Recovery Issues
• Design and verification tools
• Modeling theory
• Applications

Many ideas have been outlined in recent publications and new approaches are being proposed continuously, but few have been tested in large systems. Therefore, the timing appears to be ideal to take stock of this emerging field in a Dagstuhl seminar and work out a strategy that involves both computer scientists form academia and researchers/developers from industry. The list of invitees reflects this goal.

Creative Commons BY 3.0 DE