Improving QoS in Computational Grids through Schedule-based Approach

While Grid users are often interested in satisfaction of their Quality of Service (QoS) requirements, these cannot be satisfactory handled by commonly used queuebased approaches. This paper concentrates on the application of schedule-based methods which allow both efficient handling of QoS requirements as well as traditional machine usage objective. An incremental application of our methods reacting on dynamic character of the problem allows to achieve reasonable runtime. Even more, we show that the schedule-based methods significantly outperform queue-based approaches by means of the weighted machine usage reflecting heterogeneity of the resources common for the Grid environments. A new formalized description of two schedule-based methods is introduced to schedule dynamically arriving jobs onto the machines of the computational Grid. Earliest Gap—Earlier Deadline First (EG-EDF) policy fills the earliest gaps (EG) in the known schedule with newly arriving jobs, incrementally building a new schedule. If no gap for a coming job is available EDF policy places the new job into the existing schedule. Tabu search algorithm is used to further optimize the schedule. It moves selected jobs into the earliest suitable gaps again. Proposed methods are compared with some of the most common queue-based scheduling algorithms like FCFS (First Come First Served), EASY backfilling, and Flex-