A framework for dynamic scheduling based on quality of service metrics

In this thesis the scheduling process in production environments is improved through the development of a framework that implements a distributed dynamic scheduling methodology based on quality of service. Such methodology takes into account contingency, priority fluctuations and incorporates a scheduling algorithm referred to as Quality of Service-based Maximum Urgency First (QB-MUF) algorithm. The QB-MUF algorithm gives high priority to jobs with low probability of failing according to suitable failure probabilities for particular application environments. The contribution of this research is related to the use of quality of service metrics, calculated from the job meta-data, as part of the information used in the decision making process of the scheduling.; To validate the dynamic scheduling framework, two study cases are considered. First a simplified model of the digital publishing workflow is build upon the framework in order to observe the behavior of the proposed scheduling strategy. Second, a grid environment where resources are connected via two-level hierarchical networks is simulated. In this case, the first level is a wide area network connecting local area networks at the second level. The implemented model represents the composition of two, wide-area distributed, image operators providing treatment of data images.; Experimental results, show that the QB-MUF algorithm outperforms traditional scheduling strategies such as the Minimum Laxity First and the First In First Out algorithms.