Multi‐criteria workflow scheduling on clouds under deadline and budget constraints. (18th February 2019)

• Record Type: Journal Article
• Authors: Kalra, Mala; Singh, Sarbjeet
• Is Part Of: Concurrency and computation. Volume 31:Number 17(2019)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2019-02-18
• Subjects: budget constraints -- cloud computing -- deadline constraints -- Genetic Algorithm -- Intelligent Water Drops algorithm -- workflow scheduling; Parallel processing (Electronic computers) -- Periodicals; Parallel computers -- Periodicals; 004.35
• DOI: 10.1002/cpe.5193
• Languages: English
• ISSNs: 1532-0626

Abstract

Summary: Cloud Computing is increasingly favored by scientists for executing large‐scale workflow applications due to its appealing features such as availability of huge infrastructure, low operating costs, easy accessibility, and high scalability. Moreover, the resources in cloud are provisioned as per user's requirement following pay‐per‐use model. Scheduling workflows in cloud requires orchestration of tasks onto heterogeneous resources while complying with the task dependencies as well as user's Quality of Service (QoS) demands. The problem becomes challenging if there are multiple QoS requirements which conflict with each other. Recent studies attempted to minimize the makespan and execution cost under deadline and/or budget constraints. However, the failure of resources in cloud poses a serious threat to successful execution of these applications. Hence, it becomes essential to consider reliability of resources in addition to other parameters. This paper presents a workflow scheduling policy which minimizes makespan and execution cost while maximizing the reliability of executing workflows under user specified deadline and budget restraints. We have devised a hybrid of Intelligent Water Drops algorithm and Genetic Algorithm (IWD‐GA) to accomplish the desired goal. It provides a set of non‐dominated solutions with different makespan, cost, and reliability to offer users more flexibility in selecting a solution as per their preferences. The evaluation of the proposed algorithm has been carried based on two performance metrics, ie, hypervolume and two set coverage. The simulation results thus obtained using four diverse scientific workflows confirm that IWD‐GA outperforms non‐dominated sort genetic algorithm (NSGA‐II) and hybrid particle swarm optimization (HPSO) with regard to accuracy and diversity.