Research On Task Scheduling Strategy Under Multi-cloud Environment

With the rapid development of science and technology,cloud computing technology has been widely used for personal and commercial purposes.Cloud computing is one of the most successful technologies to provide on-demand services over the Internet.It brings extremely convenient services to customers and enterprises due to its advantages such as super-large scale,virtualization,high reliability,versatility,high scalability,and pay-on-demand.However,with the development of content sharing and collaborative computing services such as online social networks and scientific workflows,a large amount of data has been generated.The unprecedented data explosion has brought huge value and challenges to traditional data storage or processing technologies.The capacity of the cloud data center is not unlimited and cannot meet the business needs during peak hours.In order to handle such a huge amount of data,a multi-cloud system that integrates multiple clouds to provide unified services in a collaborative manner has been introduced.Taking the multi-cloud environment as the research background,this paper constructs a multi-objective and many-objective scheduling model for task scheduling problems,and designs an efficient multiobjective optimization algorithm and many-objective optimization algorithm to address the models.The main work of this paper is introduced as follows:(1)Aiming at the task scheduling problem in a multi-cloud environment,a multi-objective task scheduling model in a multi-cloud environment is build.The multi-cloud environment can combine multiple cloud providers to provide customers with unified services in a collaborative manner,which can not only avoid vendor lock-in issues,but also use the unique advantages of each vendor to meet the diverse needs of customers.However,in such a heterogeneous multicloud environment,task scheduling is more challenging than task scheduling used in a single cloud environment.In order to address the scheduling problem in a multi-cloud environment,this paper first designs a scheduling model with the objective of maximum completion time and cost.At the same time,to address the model,this paper carefully analyzes and summarizes the classic multi-objective optimization algorithm and designs experiments to address the model.(2)Since the multi-objective model only starts from the customer's point of view,takes time and cost as the objective to improve scheduling performance,and cannot fully describe the scheduling problem.In other words,these objectives cannot satisfy the needs of customers and cloud providers at the same time.For cloud providers,how to reduce machine energy consumption,improve resource utilization,and reduce machine load are very important service quality indicators.Therefore,this paper is dedicated to solving the task scheduling problem in a multi-cloud environment from the perspective of many-objectives,so as to alleviate the problem of low data processing efficiency.We introduce the idea of many-objective modeling under the multi-cloud environment,and construct a many-objective task scheduling model that includes 6 objectives from the customer's perspective(time and cost)and the provider's perspective(load balancing,energy consumption,resource utilization,and cloud throughput).And we adopt advanced many-objective optimization algorithm to address the model.(3)To further improve the efficiency of model solving,this paper designs a unified integration of many-objective optimization algorithm based on temporary offspring.The research of many-objective optimization algorithm can provide effective tools and paradigms for solving most social engineering optimization problems.However,complex and changeable optimization problems pose a challenge to the performance of the algorithm.How to make the algorithm suitable for solving different types of optimization problems and improve the efficiency of the solution has become an urgent problem to be addressed.Since the many-objective optimization algorithm is a population-based optimization algorithm,the quality of the population directly determines the efficiency of the algorithm optimization,so the convergence and diversity of the population are regarded as two important considerations.In addition,we also note the impact of the randomness of the initial population on the performance of the algorithm.Therefore,this paper will design a many-objective optimization algorithm from two aspects of temporary offspring optimization and selection operator,and propose a unified integrated many-objective optimization algorithm based on temporary offspring.The algorithm is compared with other advanced manyobjective optimization algorithms on the DTLZ and Ma OP test suites,and the statistical method Wilcoxon test is used for data analysis.Experimental results show that the proposed algorithm has optimal performance and is more suitable for solving many-objective optimization problems.At the same time,the algorithm is used to address the many-objective scheduling model proposed in Chapter 3,and the result is better than other algorithms,further verifying the performance of the algorithm,and giving the optimal task scheduling scheme,thereby improving the security of the system.This work provides a new idea for solving data processing problems.