ReoService: Coordination Modeling Tool

Coordination in SOA addresses dynamic topologies of interactions among services. Most efforts up to now have been focused on statically defined composition of services, e.g., using BPEL. To the best of our knowledge, there are no serious means to address the issues of dynamic coordination to accommodate continuously changing requirements. While BPEL is a powerful standard for service composition, it lacks support for typical coordination constraints, like synchronisation, mutual exclusion, and context-dependency. In this paper we present ReoService, which is a modeling tool for coordinating business processes. ReoService is based on Reo [2] – a general framework for coordinating components in distributed systems. Reo is a channel-based exogenous coordination language wherein complex coordinators, called connectors, are compositionally built out of simpler ones. The simplest connectors are a set of user-defined communication channels with well-defined behavior. The emphasis in this model is on connectors, not on the services to connect. In this sense, ReoService acts as a “glue” language that interconnects and coordinates services in a distributed business process. The Reo coordination tool is developed to aid the process designers who are interested in complex coordination scenarios. The ReoService and its underlying Reo framework are implemented in Java as a set of plug-ins [1] on top of the Eclipse platform (www.eclipse.org). Currently the framework consists of the following parts: (i) graphical editors, supporting the most common service and communication channel types; (ii) a simulation plug-in, that generates flash animated simulations on the fly; (iii) BPEL converter, that allows conversion of Reo connectors to BPEL and vice versa; (iv) Java code generation plug-in, as an alternative to BPEL, represents service coordination model as a set of Java classes; (v) validation plug-in, that performs model checking over coordinations represented as constraint automata. We now describe the Reo framework architecture that is shown in Figure 1. The central part of the framework is a visual editor for Reo connectors. It represents the actual coordination model with services and communication channels. The developed tool also allows us to represent Reo in terms of constraint automata [4]–an alternative behavioral model. This is useful if additional validation based on model checking techniques [6] is required. Q-Automata [5] is used if