Strollo, Daniele (2009) Designing and experimenting coordination primitives for service oriented computing. Advisor: Ferrari, Prof. Gianluigi. Coadvisor: Tuosto, Dr. Emilio . pp. 146. [IMT PhD Thesis]
Strollo_phdthesis.pdf - Published Version
Available under License Creative Commons Attribution No Derivatives.
Download (4MB) | Preview
Service Oriented Architecture (SOA) and Web Services (WS) are becoming a widely accepted device for designing and implementing distributed systems. SOAs have given an important contribution to software engineering providing a model where applications are defined by assembling together certain functionalities, called services, possibly provided by remote suppliers. The characterizing issue of SOAs consists of defining common principles to make services accessible and usable regardless their execution context. Nevertheless, the architectural specification is far from giving a complete reference application model on which systems should rely on. The specification just includes principles for achieving interoperability and reusability of services; other aspects are left to the implementing platforms. As a consequence, it is understood how services are specified in isolation and how their functionalities are made available to the requesters, but the definition of languages for describing service composition are far from being widely accepted and reveals to be an impelling challenge. In the last years, several solutions have been proposed for describing aggregated services. However, they often lack a formally defined semantics. Moreover, these solutions are often specific for a platform (e.g. WSs) and are difficult to adapt to other platforms since they rely on low level assumptions that are out of the SOA specifications. This thesis aims at providing new methodologies for implementing the coordination of services. Our framework proposes to be flexible enough to support high level languages and to provide reliable tools for testing correctness of implementation. Our approach relies on a formal model that takes the form of a process calculus specifically designed to deal with services and their coordination. The process calculus has been the main tool driving the specification issues as well the implementation issues. Indeed, it acts as a bridge between the high level specification language and the run-time environment. A distinguished feature of our proposal is that our formal model, i.e. the process calculus, describes distributed processes relying on an event notification mechanism as machinery for interactions. Services are represented by certain components that embody local computations and react to changes of the overall environment in which they are involved. The adoption of event notification results particularly fashionable for tackling service coordination. The principles studied at specification level are from one side understood within a theoretical framework that provides instruments for checking correctness of interaction policies and from the other side offers the core model for implementing and experimenting a programming middleware.
|Item Type:||IMT PhD Thesis|
|Subjects:||Q Science > QA Mathematics > QA75 Electronic computers. Computer science|
|PhD Course:||Computer Science and Engineering|
|Date Deposited:||06 Jul 2012 09:00|
Actions (login required)