Title: Adaptive Workflow Composition in Service-based Systems (Aanpasbare workflowcompositie in service-gebaseerde systemen)
Other Titles: Adaptive Workflow Composition in Service-based Systems
Authors: Geebelen, Kristof
Issue Date: 7-Sep-2012
Abstract: The paradigm of service-oriented computing (SOC) has received a lot of attention over the last years and has changed the way software systems are designed. The promise of SOC is to provide flexibility and agility on the level of systems development. At the heart of service-based systems are services that provide platform- and technology-independent, computational elements that can be composed into loosely coupled networks of collaborating applications in order to reflect an organisation's business-level objectives. A popular approach for defining such business flows as a cooperation of services is workflow (or service) composition. The shift towards SOC has also impacted the delivery and consumption of software systems. Recently, cloud computing has become an emerging mechanism that fits into this evolution. The combination of fast expanding technologies with increasing competitiveness drives more and more companies to the outsourcing of parts of their business IT operations to third-party cloud providers. This enables them to take advantage of the cloud provider's expertise and to reduce their cost by sharing resources to exploit economies of scale. In the cloud, service-based systems can be deployed as Software-as-a-Service (SaaS) applications that can be purchased and consumed remotely over the internet.Due to increasing user demands and evolving requirements these service-based systems should be able to adapt to both dynamic changes in operational requirements and environmental conditions, while providing predictable behavior with respect to service qualities such as response time, availability, scalability, and security. State-of-the-art industry standards in workflow composition, however, exhibit major limitations regarding modularity and flexibility to support complex and highly dynamic service compositions. Hence, in this context three interesting research challenges appear that are addressed in this dissertation. First, traditional service composition solutions should be able to enforce policy (or rule) specifications that describe how the workflow composition must adapt to changing requirements and circumstances. Second, an important aspect of adaptive service composition is that it requires the capability to dynamically change the levels of quality of service (QoS) to satisfy customers demand by selecting the appropriate participating services. Third, with the trend of cloud computing, cloud providers want to make full use of economies of scale by hosting their services following a multi-tenant model, where a single application is used to serve multiple customers (called tenants). In such a multi-tenant setting, the cloud provider should be able to tailor its business processes to meet the functional and non-functional requirements of each tenant.This dissertation delivers four core contributions. First, we present a realistic application that has been developed using state-of-the-art technologies in service-oriented computing. This application is used as a practical case-study to expose some key limitations in current industry standards and technologies that motivate our remaining contributions. Second, we propose a portable framework for the enforcement of dynamic adaptation policies in business processes. The framework is inspired by the model-view-controller (MVC) pattern, commonly used for adding dynamism to web pages. Third, we present a theoretical approach to effectively deal with the composition of services that require certain levels of quality. As actual QoS support of participating services changes over time, the service composition problem must be treated as a decision problem under uncertainty. Therefore, we have developed an algorithm for predicting whether the QoS of a service composition execution will be compliant with a service level agreement (SLA) between a customer and the service (composition) provider. Fourth, we capitalize on the shift from supply to demand driven processes by proposing a middleware that provides the mechanisms to perform requirement-driven adaptation of shared process templates in a multi-tenant SaaS environment.We validate our work in several respects. To demonstrate the feasibility of our work, the proposed ideas have been prototyped. To measure the expressiveness of our adaptive workflow approach, we analyze its customization support by means of an extensive classification of change patterns and change support features. To analyze the performance of our prediction algorithm, used for selecting appropriate participating services in a composition, we introduce two performance indicators for comparing different QoS prediction algorithms. Our validation on both real as well as simulated QoS data shows that the proposed algorithms outperform existing approaches. To evaluate the effectiveness of tenant-specific customization in the context of Software as a Service in practice, we apply our approach on a case-study in the healthcare domain. Our measures confirm that the middleware improves the QoS of a composition with an acceptable performance overhead.
Publication status: published
KU Leuven publication type: TH
Appears in Collections:Informatics Section

Files in This Item:
File Status SizeFormat
thesis-kristofgeebelen.pdf Published 8570KbAdobe PDFView/Open


All items in Lirias are protected by copyright, with all rights reserved.