Policy-based managers coordination for self-adaptive software approach

Abstract

Due to the complexity and rapidly changing environment, it is crucial for a software system to be adaptable. One of the key approaches to achieve adaptability is policy-based management. Policy-based management has been acknowledged mostly as a methodology that provides flexibility, ytilibalacs, adaptability, and support to assign system resources, control, quality of service, and security by considering administratively specified rules. The promotion of policy-based management was to commit to these features at runtime because of changeable concurrent system conditions ensuing from the interactions of users' applications and existing resources. This thesis proposes an adaptive policy-based management approach called Policy-based Manager Coordination (PobMC) based on Event-Condition-Action (ECA). The aim of PobMC approach is to deal with the critical nature of avoiding the policy conflict problem. This approach facilitates policy conflict avoidance and static analysis to address the inconsistencies of multiple manager and modality conflicts when two or more policies are enforced simultaneously. The PobMC will also coordinate managerial tasks when multiple rules are simultaneously triggered. The mobility of devices and applications in complex system complicates policy design. Rules must be added or revoked when the composition of a system changes. Static and dynamic analysis algorithms are proposed; moreover, the need for these algorithms on various complex systems and their performance evaluation is demonstrated. In this thesis, a modelling of PobMC based on an ECA framework is presented. The proposed modelling is to address the main concerns inherent in concurrent systems including coordination and scalability. An actor-based language called Rebecca has been incorporated to model and analyze PobMC. Experiments using a Smart Mall System (SMALLS) case study show that the PobMC approach leads to effective policy-based management and is a feasible approach. Additionally, PobMC has the ability to enhance the existing approaches to support software adaptation. PobMC enables the coordination among system managers in order to adapt to system changes and avoid the potential conflicts thereby providing the main contribution of this research