|Thesis abstract: |
Modern information systems often operate in dynamic environment where resources are inconsistent, user needs frequently change and operating context constantly varies. This dynamic property makes the system easily become prone of errors and vulnerable to malicious attacks; therefore triggers the need to have a system that is able to self-aware environmental changes and self-adapt to changes in order to maintain its functional and non-functional requirements. While there are several proposals in literature focusing on designing systems that are self-managed, they are not enough to capture changes, which are often unpredictable, introduced at runtime in actual execution context; hence difficult to achieve a satisfied level of self-adaptation. In my work, I will propose runtime adaptation mechanisms for information systems, in particular for service-based business processes, taking into account data dependency and task dependency among different processes. Initially, I will exploit different quantitative analysis methods to model the system in such a way that allows understanding system behaviors and performance, base on that to predict its future behaviors and trigger adaptation actions at runtime if necessary. Defining the set of adaptation actions for the system to react is certainly a part of the work. In experimental part, I will validate the proposed mechanisms by applying them to the issue of energy efficiency in information systems; or fault tolerance system where system has self-healing property.