Implementing a RosettaNet Business-to-Business Integration Platform Using J2EE and Web Services

Abstract

The need for connecting information systems of collaborating organizations has become increasingly common. Significant advantages, such as increased speed, efficiency, and reliability, can be obtained by automating inter-organizational business processes. To achieve this, business-to-business integration, i.e. facilitating interoperation of disparate information systems of different organizations, must be performed. E-business frameworks are generic solutions for performing such integration. RosettaNet is an industry consortium that maintains an e-business framework that specifies inter-organizational business processes for multiple industries. Process specifications include messages that are exchanged between organizations, and related messaging choreography. RosettaNet Implementation Framework (RNIF) is the messaging portion of the RosettaNet framework, specifying how messages are exchanged.

The purpose of this work is to gather experience on implementing RNIF with currently available tools. Research questions include how can RNIF be implemented in practice, how suitable are current tools and what could be improved in them, how much effort is required, what level of performance can be obtained, and are there any interoperability problems with different RNIF implementations.

To answer these questions, a prototype system is developed using J2EE (Java 2 platform Enterprise Edition) and web service technologies. The prototype is a middleware system that provides RNIF functionality, on top of which RNIF-enabled applications can be constructed with less effort than creating an equivalent RNIF implementation from scratch. The prototype has two high-level goals. The first is to create a learning platform that students can use to practice RosettaNet-based integration: the platform provides RNIF-compliant messaging functionality, and students are free to concentrate on higher-level issues, such as semantics of messages that are exchanged. The second goals it to create a realistic RNIF software implementation, providing services that are assumed useful in a typical business-to business integration scenario, to gather experience on implementing such systems.

The prototype is tested in laboratory environment. Testing is done for RNIF-compliance, interoperability with a commercial RNIF implementation, and performance. In addition, a student group applies the prototype to implement an integration system for a simple RosettaNet-based integration case, to test the prototype's practical usability and ease of use.

The results show one possible way to implement RNIF in practice. The tools were suitable for the task, but some issues exist: some tools did not function perfectly and some improvement suggestions could be made. Performance level of processing in order of one RNIF message per second is relatively easily attainable; current tools do place limitations on how large messages can be transferred. Interoperability problems were detected: a commercial RNIF implementation produced invalid RNIF messages and some details in related specifications leave room for interpretation.