| With the emergence of applications and services in the context of the big data era,cloud computing plays an important supporting role and has rapidly developed.In this context,the introduction of cloud workflow systems has brought convenience to users for the purpose of meeting their emblematic requirements in multifarious exclusive scenarios.Cloud workflow systems can abstractly define complex workflow processes,establish data structure relationships for large and complex scientific computing processes and commercial industrial application processes,and submit applications such as earthquake prediction,graphics and image processing,and genome sequencing to the cloud for processing.Nevertheless,cloud workflow scheduling not only needs to study problems directly related to service quality such as makespan and economic costs,but also needs to study the dependence relations between various tasks within the workflow system.Its particular sequential process for executing tasks and the intermediate results between precursor tasks and successor tasks are considerable factors during scheduling.This article conducts research on workflow scheduling in cloud data centers,with a focus on energy saving technologies in data centers,workflow scheduling algorithms,especially meta-heuristic optimization algorithms and strategies for Multi-Objective Optimization.Unpacking the present problems in current research and in the light of the characteristics of workflow,conducted research on workflow scheduling based on ALO(Ant Lion Optimizer).The main goal is to reduce the energy consumption of cloud data centers by designing reasonable task scheduling strategies while ensuring the quality of service(reducing the makespan and economic costs).The specific contributions are as follows:(1)Research on workflow scheduling based on enhanced ant lion optimization algorithm.Since the workflow scheduling problem is NP-hard,a new meta-heuristic algorithm,Ant Lion Optimizer,is used to design scheduling strategies.According to the characteristics of workflow tasks,a Historical Optimal Recording Mechanism and a strategy of Adjusting the Wandering Step Dynamically are proposed,which overcomes the shortcomings of traditional Ant Lion Optimizer when transferring from a continuous optimization problem to a discrete optimization problem,and can prevent the population from falling into a local optimization prematurely,achieving better convergence performance.The goal is to achieve as low energy consumption as possible while ensuring that tasks are completed execution in as little time as possible.Experimental results on five scientific workflows show that the proposed workflow scheduling strategy has relatively good performance compared to other classical methods,verifying the effectiveness of the proposed method.(2)Research on multi-objective workflow scheduling based on hybrid Ant Lion Optimizer and Fireworks Algorithm.The three objectives of minimizing makespan,reducing energy consumption and cutting down economic costs are considered comprehensively.Adopting two meta heuristic methods,Ant Lion Optimizer and Fireworks Algorithm to realize complementary advantages,which not only retain the strong exploration power of the original ALO,but also enhance local development capabilities through fireworks’ s explosion.A task scheduling strategy based on hybrid Ant Lion Optimizer and Fireworks Algorithm(HAALO)is proposed.Due to the contradictory relationship between the three objectives of makespan,energy consumption and cost.The primal problem could be modelled as a Pareto-based Multi-objective Optimization Problem(MOP)which is taking the aim of optimizing the three objectives simultaneously and obtaining a solution that approaches the overall optimization.Experimental results show that the proposed algorithm has significant effects in minimizing makespan,reducing energy consumption and cutting down economic costs compared to other multi-objective optimization algorithms. |
