Research Area:  Cloud Computing

Abstract:

As an efficient development for industrial and scientific applications, workflow technologies have received substantial attention in recent decades. To address the issue of workflow scheduling in a state-of-the-art cloud environment, based on analysis of a decentralized architecture for workflow scheduling, a dependable scheduling strategy with active replica placement (DS-ARP) is proposed in this paper. In this proposal, by analyzing control/data dependencies in a workflow, a game-theory-based active replica placement model is first developed to achieve reasonable replica placement; then, a dependable scheduling algorithm is proposed to enhance the system reliability and security. With five well-known workflow applications, CloudSim-based simulations are performed, and the analytical results are shown to demonstrate the performance of DS-ARP on an average number of initiated replicas, costs resulting from canceled replicas, makespans, deadline violation rates and resource utilization rates.

Keywords:  
Cloud Computing
Job Shop Scheduling
Scheduling Algorithms
Quality Of Service
Dynamic Scheduling
Workflow
Cloud
Control Data Dependencies
Active Replica Placement
Dependable Scheduling

Author(s) Name:  Ming Tao, Kaoru Ota, Mianxiong Dong

Publisher name:  IEEE

DOI:  10.1109/TCC.2016.2628374

Volume Information:  volume 8,(2020)

Paper Link:   https://www.computer.org/csdl/journal/cc/2020/04/07742980/13rRUxYINai