Scheduling theory maintains that there are fundamental similarities in problems of sequence that transcend the characteristics of the particular tasks to be ordered or the resources to be used. Traditionally, scheduling policies are implemented using algorithms; we study scheduling algorithms to discover the various properties of the schedules they produce. To facilitate analysis the policies are typically limited to homogeneous task sets (e.g., all periodic tasks) and consider only one or very few task attributes. In some cases the results are so attractive that the task sets of systems are made to fit the algorithm rather than using a policy more appropriate to the system. We therefore make the following observation: if scheduling policies are driven by how well they can be expressed and analyzed, then we need a more general framework for expressing scheduling policies.We introduce the Importance Abstraction as a general scheduling framework. The scheduling algorithm is invariant: choose the most important task at every point in time. Each task is described by a function, called an importance function, that profiles the task's importance to the system over time. The importance abstraction can express not only the traditional scheduling policies but a wide range of new policies based on how important individual tasks are to the system. Since the scheduling policy is described using functions rather than a single algorithm we can exploit the maturity of mathematical proof techniques when analyzing the schedule produced by the policy. Since this abstraction is applicable to any system of tasks and processors, we examine the communication subsystem as an example, and find that importance functions facilitate the expression of message discrimination policies as well as help unify scheduling across the operating system/communication subsystem domain boundary.
Cited By
- Yu Y, Ren S, Chen N and Wang X Profit and penalty aware (PP-aware) scheduling for tasks with variable task execution time Proceedings of the 2010 ACM Symposium on Applied Computing, (334-339)
- Li P, Wu H, Ravindran B and Jensen E (2006). A Utility Accrual Scheduling Algorithm for Real-Time Activities with Mutual Exclusion Resource Constraints, IEEE Transactions on Computers, 55:4, (454-469), Online publication date: 1-Apr-2006.
Index Terms
- Function-driven scheduling: a general framework for expression and analysis of scheduling
Recommendations
Global EDF-based scheduling with laxity-driven priority promotion
This paper presents an algorithm, called Earliest Deadline Critical Laxity (EDCL), for scheduling sporadic task systems on multiprocessors. EDCL is a derivative of the Earliest Deadline Zero Laxity (EDZL) algorithm. Each job is assigned a priority based ...
Priority-Driven Scheduling of Periodic Task Systems on Multiprocessors
The scheduling of systems of periodic tasks upon multiprocessor platforms is considered. Utilization-based conditions are derived for determining whether a periodic task system meets all deadlines when scheduled using the earliest deadline first ...
Due date assignments and scheduling a single machine with a general earliness/tardiness cost function
We study three different due date assignment problems in scheduling a single machine which differ from each other based upon the objective function and due date assignment method being used. Two different objective functions are considered. The first is ...