Multiple DAGs Workflow Scheduling Algorithm Based on Reinforcement Learning in Cloud Computing
To the problem of scheduling multiple DAG workflow applications with multiple priorities submitted at different times in cloud computing environment, a novel workflow scheduling algorithm based on reinforcement learning is proposed in this paper. In the w
- PDF / 376,178 Bytes
- 7 Pages / 439.37 x 666.142 pts Page_size
- 70 Downloads / 188 Views
Abstract. To the problem of scheduling multiple DAG workflow applications with multiple priorities submitted at different times in cloud computing environment, a novel workflow scheduling algorithm based on reinforcement learning is proposed in this paper. In the workflow scheduling scheme, the number of VMs in resources pool is defined as state space; the runtime of user task is defined as immediate reward, and then interactive with cloud computing environment to obtain the optimization policy. We use real cloud workflow to test the proposed scheme. Experiment results show the proposed scheme not only can solve the fairness of scheduling multiple DAGs with the same priority level submitted at different times, but also can ensure that the execution of the DAGs with higher priorities cannot be influenced by the DAGs with lower priorities. More importantly, the proposed scheme can reasonably schedule multiple DAGs with multiple priorities and improve utilization rate of resources better. Keywords: Multiple DAGs Reinforcement learning Workflow scheduling Cloud computing
1 Introduction A system that defines, creates and manages the execution of workflows through the use of software, running on one or more workflow engines, which is able to interpret the process definition, interact with workflow participants and, where required, invoke the use of IT tools and applications [1]. A workflow can usually be described using formal or informal flow diagramming techniques, showing directed flows between processing steps. Single processing steps or components of a workflow can basically be defined by three parameters [2]: Input description: the information, material and energy required to complete the step. Transformation rules, algorithms, which may be carried out by associated human roles or machines, or a combination.
© Springer Science+Business Media Singapore 2016 K. Li et al. (Eds.): ISICA 2015, CCIS 575, pp. 305–311, 2016. DOI: 10.1007/978-981-10-0356-1_31
306
D. Cui et al.
Output description: the information, material and energy produced by the step and provided as input to downstream steps. Recently years, Workflow Management System (WMS) is arranged in cloud computing environment became a new cloud computing application. Lots of system models and methods which aim at cloud workflow application or Directed Acyclic Graph (DAG) task scheduling are designed. Dornemann designed an on-demand resource provisioning for BPEL workflows using amazon’s elastic compute cloud [3]. Byun proposed cost optimized provisioning of elastic resources for application workflows [4]. They all realized the resource provisioning schemes in real Amazon EC2 cloud computing platform. But only few research works have been done in multiple DAGs workflow Scheduling problem [5–11]. In this paper, in order to solve the problem of scheduling multiple DAG workflow applications with multiple priorities submitted at different times in cloud computing environment, we proposed a multiple DAGs cloud workflow scheduling algorithm based on Reinforcement
Data Loading...
