Scheduling bag-of-tasks applications on hybrid clouds under due date constraints

Bag-of-tasks (BoT) applications consisting of multiple tasks are widely used in a variety of application fields. Different from the traditional deadline constraint in scheduling problems, a due date constraint allows the completion time of BoT applications to exceed the pre-specified due date, but would impose the tardiness penalty. In order to reduce this tardiness penalty, the customer is willing to outsource a part of tasks to public clouds despite the new cost of using public clouds' resources. In this scenario, the challenge is to optimally schedule tasks on hybrid clouds for the purpose of minimizing the total cost. This problem of scheduling BoT application on hybrid clouds under due date constraints can be formulated as an integer program. To solve this problem, we propose several heuristics by employing an effective task rescheduling strategy. Furthermore, we develop two effective acceleration methods, namely AM1 and AM2, which can be used by the rescheduling strategy for improving the efficiency. This conclusion is validated by proofs in theory. Extensive simulations are performed to justify the effectiveness and efficiency of the proposed scheduling heuristics. Experimental results demonstrate the performance of the task rescheduling strategy in terms of rescheduling quality and computational efficiency.