Abstract
The Integration Rules (IRules) project at Arizona State University
(http://www.eas.asu.edu/~irules) is developing a declarative event-based
approach to component integration. Integration rules are based on the concept
of active database rules, providing an active approach for specifying event-
driven activity in a distributed environment. The IRules project consists of a
knowledge model that specifies the IRules Definition Language and an execution
model that supports integration rule execution. This research focuses on the
execution model and the architectural design parts of the IRules project. The
main objective of this research is to develop a distributed execution
environment for using integration rules in the integration of black-box
components. In particular, this research will investigate the design of an
architecture that supports the IRules semantic framework, the development of
an execution model for rule and transaction processing, and the design of a
rule processing algorithm for coordinating the execution of integration rules.
This research will combine the distributed computing framework of Jini, the
asynchronous event notification mechanism of the Java Message Service (JMS),
and the distributed blocking access functionality of JavaSpaces to support
active rule processing in a distributed environment. The limitations of the
underlying Enterprise JavaBeans (EJB) component model pose transaction
processing challenges for the integration process. This research will develop
a suitable transaction model and processing logic to overcome the limitations
of the underlying EJB component model. Furthermore, the architectural design
will allow an easy extension of the system to accommodate other component
models. This research is expected to contribute to nested rule and transaction
processing for active rules that have not been previously addressed in
distributed rule processing environments. The development of the IRules
execution environment will also contribute to the use of distributed rule-
based techniques for eventdriven component integration.