Investigating Instrumentation Techniques for ESB Runtime Verification

Abstract

Abstract. Enterprise Service Buses (ESBs) are highly-dynamic component platforms that are hard to test for correctness because their connected components may not necessarily be present prior to deployment. Runtime Verification (RV) is a potential solution towards ascertaining correctness of an ESB, by checking the ESB's execution at runtime, and detecting any deviations from the expected behaviour. A crucial aspect impinging upon the feasibility of this verification approach is the runtime overheads introduced, which may have adverse effects on the execution of the ESB system being monitored. In turn, one factor that bears a major effect on such overheads is the instrumentation mechanism adopted by the RV setup. In this paper we identify three likely (but substantially different) ESB instrumentation mechanisms, detail their implementation over a widely-used ESB platform, assess them qualitatively, and empirically evaluate the runtime overheads introduced by these mechanisms.

Runtime Verification for Decentralised and Distributed Systems

“…They provide handles that allow the specifier to dictate whether a property is to be runtime-checked locally, as inlined code within the monitored system, or remotely via an independent monitoring unit located on a separate machine. In separate work Colombo et al [32] investigate various instrumentation techniques in Enterprise-Service Bus (ESB) distributed architectures, so as to determine which of them lead to lower monitoring overheads.…”

Runtime Verification using LARVA

“…As a means of exploring ways in which runtime verification could be applied in this context, Larva has been instrumented in several ways together with the Mule Enterprise Service Bus [10].…”