Applying runtime verification techniques to enterprise service bus architectures

Abstract

The increased usage of distributed systems has led to the problem of integration of IT systems that communicate via different protocols. In such setting, it is also typical for these components to be added/removed from the system at runtime, depending on requirements. These two characteristics give rise to integration challenges where systems should be able to communicate seamlessly with each other whilst being able to be easily integrated with an existing distributed system with minimal impact. An ESB (Enterprise Service Bus) is a tool which attempts to address the above issues by integrating systems in a bus-like architecture where a central bus enables components to communicate via messages. This arrangement, transparently handling complex messaging functions such as transformation and routing (via routing patterns), enables the user to focus on the business logic, abstracting away communication concerns [6]. Despite facilitating orchestration of distributed components in a scalable manner, current ESBs provide little support for correctness guarantees of the overall system logic e.g. a booking component may only confirm the booking once the bank component has verified the payment details, and the airline component confirms that the dates specified are permissible. Popular techniques for ensuring correctness, such as testing and model checking are not ideal for verifying the correctness of ESB systems due to the latter’s highly dynamic nature. For this reason, we propose to apply runtime verification techniques, promising a scalable approach to checking all observed behaviour under any runtime circumstance — no matter how unpredictable this may be.