A specification and analysis framework for provenance awareness of service compositions

• Main Author: Zerva, Paraskevi
• Other Authors: Miles, Simon ; Zschaler, Steffen
• Published: King's College London (University of London) 2018
• Subjects: 004
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.733442

The Service-Oriented Computing (SOC) paradigm, realizing a software design philosophy and architecture - the service-oriented architecture (SOA) - provides the means to develop service-oriented systems by exploiting and composing loosely coupled services. As service-oriented systems become increasingly large-scale and infrastructure-heterogeneous, their execution leads to excessive data production stressing the need for service providers to exhibit accountability about the systems' qualities and actions. Accountability requires that systems faithfully document their execution, being able to answer questions about how the data was produced and processed, referred to as data provenance. Provenance awareness is the functionality providing access to provenance by allowing users to query data about past processes and answer provenance questions. The SOC paradigm promotes composability. For a composite service, the independent audits of atomic services do not simply compose to a connected queryable picture of the provenance across the end-to-end pathways of a composition's execution. Taking into account the inherent complexity of service discovery, selection and dynamic composition aspects of SOC, we realize that we need to carefully design for provenance awareness support by modeling explicitly the provenance data structures and infrastructure independently of specific applications. This thesis proposes a set of formal models, analysis methods, and tools to address this need. We contribute a formal specification and analysis framework for provenance-awareness of service compositions introducing: a faceted classification of provenance questions to formally express provenance requirements acting as analytical metrics for provenance-awareness, a provenance data model capturing the provenance structures of service compositions, a template provenance infrastructure metamodel allowing one to design the provenance infrastructure of atomic and composite services, and a modeling and analysis environment that verifies through simulation and analysis whether the provenance infrastructure system design satisfies a set of provenance requirements. As a proof of concept, we provide a prototype tool implementation of our framework.