| |
Freitas, A. L., Parlavantzas, N., & Pazat, J. - L. (2010). A QoS Assurance Framework for Distributed Infrastructures. In MONA’10: Proceedings of The Third International Workshop on Monitoring, Adaptation and Beyond. Acm.
Abstract: Enforcing SLAs (Service Level Agreements) for services deployed on large-scale distributed infrastructures, such as grids and clouds, is complex owing to fluctuating customer demand and unpredictable resource availability. Current systems either address specific application domains or fail to provide a complete QoS assurance solution. This work proposes a generic framework to assist service providers in enforcing quality properties in distributed environments. The framework provides a rich set of QoS management functions, including negotiation, translation, and resource provisioning. Importantly, the framework supports dynamic adaptation; that is, it automatically modifies service behavior and resource usage in order to maintain agreed service levels while satisfying service provider-specific constraints. We have implemented an initial prototype in a grid environment and demonstrated its effectiveness in minimizing SLA violations.
Keywords: Service-Oriented Architecture, QoS assurance, Self-Adaptation
|
|
Freitas, A. L., & Pazat, J. (2010). A Self-Adaptable Approach for Easing the Development of Grid-Oriented Services. In IEEE 10th International Conference on Computer and Information Technology (CIT 2010) (pp. 76–82).
Abstract: The Service-Oriented Architecture (SOA) leverages the service abstraction to enable the development of modular, loose-coupled and distributed applications. In order to use such an architecture, service-based applications directly rely on services or compose them for conceiving new functionalities. In spite of these capabilities, they do not support the development of services which need high-performance computing. Grid computing offers an infrastructure for high-performance computing which is based on the sharing of distributed, low-cost and heterogeneous resources in large-scale. Thus, grids can be used to satisfy these high-performance service requirements. This work aims at easing the development of grid-oriented services. The iPOJO service-component model is used to propose an architecture that automatically manages job submission for services. This architecture is based on Dynaco (Adaptation for Components) and the XtreemOS grid operating system.
|
|
Freitas, A. L., Parlavantzas, N., & Pazat, J. - L. (2011). Cost Reduction through SLA-driven Self-Management. In 9th IEEE European Conference on Web Services, ECOWS 2011 (pp. 117–124). IEEE.
Abstract: A main challenge for service providers is managing service-level agreements (SLAs) with their customers while satisfying their business objectives, such as maximizing profits. Most current systems fail to consider business objectives and thus to provide a complete SLA management solution. This work proposes an SLA-driven management solution that aims to maximize the provider’s profit by reducing resource costs as well as fines owning to SLA violations. Specifically, this work proposes a framework that comprises multiple, configurable control loops and supports automatically adjusting service configurations and resource usage in order to maintain SLAs in the most cost-effective way. The framework targets services implemented on top of large-scale distributed infrastructures, such as clouds. Experimental results demonstrate its effectiveness in maintaining SLAs while reducing provider costs.
|
|
Freitas, A. L., Pazat, J. - L., & Parlavantzas, N. (2010). Ensuring Resource-Level Quality for Services on Grids. In Proceedings of the 2010 6th World Congress on Services (pp. 168–169). Services ’10. Washington, DC, USA: IEEE Computer Society.
Abstract: Service-Oriented Architectures address the development of distributed and dynamic service-based applications. Due to the dynamics of their environments, services should be self-adaptable in order to maintain agreed resource-level qualities. To support building such services, this work proposes the Self-Adaptable Service Execution Manager (SASEM), responsible for monitoring and controlling the service execution on grid resources in order to prevent SLA violations. SASEM builds on the Dynaco adaptation model and on an enhanced grid monitoring and actuation system, implemented using the XtreemOS grid operating system.
|
|
S. Lane, A. B. and I. R. (2011). SOAdapt: A Process Reference Model for Developing Adaptable Service-Based Applications. Information and Software Technology Journal, .
Abstract: Context: The loose coupling of services and Service-Based Applications (SBAs) have made them the ideal platform for context-based run-time adaptation. There has been a lot of research into implementation techniques for adapting SBAs, without much effort focused on the software process required to guide the adaptation.
Objective: This paper aims to bridge that gap by providing an empirically grounded software process model that can be used by software practitioners who want to build adaptable SBAs. The process model will focus only on the adaptation specific issues.
Method: The process model presented in this paper is based on data collected through interviews with 10 practitioners occupying various roles within eight different companies. The data was analyzed using qualitative data analysis techniques. We used the output to develop a set of activities, tasks, stakeholders
and artifacts that were used to construct the process model.
Results: The outcome of the data analysis process was a process model identifying nine sets of adaptation process attributes. These can be used in conjunction with an organisation’s existing development lifec-cycle
or another reference life-cycle.
Conclusion: The process model developed in this paper provides a solid reference for practitioners who
are planning to develop adaptable SBAs. It has advantages over similar approaches in that it focuses
on software process rather than the specific adaptation mechanism implementation techniques.
|
|
|
