Abstract
The provision of resources and services for scientific workflow applications using a multi-cloud architecture and a pay-per-use rule has recently gained popularity within the cloud computing research domain. This is because workflow applications are computation intensive. Most of the existing studies on workflow scheduling in the cloud mainly focus on finding an ideal makespan or cost. Nevertheless, there are other important quality of service metrics that are of critical concern in workflow scheduling such as reliability and resource utilization. In this respect, this paper proposes a new multi-objective scheduling algorithm with Fuzzy resource utilization (FR-MOS) for scheduling scientific workflow based on particle swarm optimization (PSO) method. The algorithm minimizes cost and makespan while considering reliability constraint. The coding scheme jointly considers task execution location and data transportation order. Simulation experiments reveal that FR-MOS outperforms the basic MOS over the PSO algorithm.
Highlights
Nowadays, cloud computing technology has become one of the most prominent technologies that provide computing resources to end users
All the virtual machines (VMs) provided by different clouds are designed to constitute the entire search space in which the particle swarm optimization (PSO) searches the execution location and the particles move to selected VMs by tasks
This study focused on virtual machines (VMs) of three commercial cloud providers (Amazon EC2, Microsoft Azure and Google compute engine)
Summary
Cloud computing technology has become one of the most prominent technologies that provide computing resources to end users. The main aim of the proposed algorithm is to minimize cost and makespan, taking into account reliability constraints In this regard, the scientific workflow schedule accounts for the following issues: (1) the IaaS cloud platform to be selected; (2) the type of VM to be assigned to the tasks; and (3) the order of tasks that should be utilized for data transmission. The scientific workflow schedule accounts for the following issues: (1) the IaaS cloud platform to be selected; (2) the type of VM to be assigned to the tasks; and (3) the order of tasks that should be utilized for data transmission To address these issues, the FR-MOS algorithm deploys particle swarm optimization (PSO) and considers task orders and task execution location in its coding strategy. Simulation results show that the FR-MOS algorithm gives better results than MOS algorithm [21] when deployed for scheduling different scientific real-world workflow
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
