An Adaptive Heuristic for Scheduling Dynamic and Fuzzy Jobs on Elastic Clouds

Jie Zhu, Han Liu, Haiping Huang

Abstract: We consider a special workflow scheduling problem in a cloud-based workflow management system in which tasks have the fuzzy processing times, fuzzy due dates, linearly dependency, stochastic arriving times and executed on elastic cloud resources with multiple price structurings. A triangle fuzzy number based problem model is formulated. Two metrics are explored: the cost and the satisfaction degree. In order to improve the solutions, the price structuring of VMs can be adjusted according to their real utilization. An adaptive and iterated heuristic framework is presented to schedule tasks event by event which mainly consists of the task collecting and the fuzzy task scheduling. The proposed methods are calibrated through detailed designs of experiments and sound statistical techniques. The proposed algorithm is compared to existing methods for related problems. Experimental results show that the proposal is robust and effective for the problems under study.