Keep it small, keep it real: efficient run-time verification of Web service compositions

Service compositions leverage remote services to deliver added-value distributed applications. Since services are administered and run by independent parties, the governance of service compositions is intrinsically decentralized and services may evolve independently over time. In this context, pre-deployment verification can only provide limited guarantees, while continuous run-time verification is needed to probe and guarantee the correctness of compositions at run time. This paper addresses the issue of efficiency in the run-time verification of service compositions described in BPEL. It considers an existing monitoring approach based on ALBERT, which is a temporal logic language suitable for asserting both functional and non-functional prop- erties, and shows how to obtain the efficient run-time verification of ALBERT formulae. The paper introduces an operational semantics for ALBERT through an extension of alternating automata, and explains how to optimize it to produce smarter, and thus more efficient, encodings of defined formulae. The optimized operational semantics can then be the basis for an efficient implementation of the run-time verification framework.