QoS monitoring of soft contracts for transaction based Web services orchestrations⋆

Abstract. Web services orchestrations and choreographies require es-tablishing Quality of Service (QoS) contracts with the user. This is achieved by performing QoS composition, based on contracts established between the orchestrator and the Web services called in the orchestration. Typical QoS parameters include maximum query throughput, response time, security, and validity of the response. Usually, QoS contracts are stated in the form of hard guarantees (e.g., response time always less than 5 msec). However, experiments and measurements from existing Web services show evidence that soft guarantees, not hard, should be stated instead (e.g., response time less than 5 msec in 95 % of the cases). In a previous work [1] we have proposed using soft contracts, by taking a probabilistic approach. Contracts are characterized by means of a set of selected quantiles of probability distributions for QoS parameters. We showed how to compose such contracts, to yield a global QoS (proba-bilistic) contract for the orchestration. Our approach is supported by the TOrQuE tool, that performs probabilistic contract composition, au-tomatically, from orchestration specification and QoS contracts with the Web services called by the orchestration. In this work in progress, we develop monitoring techniques for detecting the violation of such soft probabilistic contracts. Monitoring of the called services is performed by the orchestration. We propose two simple sta-tistical methods for monitoring the called services for contract violation.