| Since the development of computational science,great achievements have been made in scientific research,engineering experiments,and research and development of military weapons.In recent years,the research and development of high-performance computing has been highly valued by countries all over the world and has risen to become an important strategic goal of the country.The high-performance computing cloud platform is an efficient service platform that combines high-performance computing with cloud computing technology.It integrates the underlying resources through virtualization technology to provide users with high-performance computing services.This thesis is supported by a high-performance computing cloud platform built by a research institute of China Aerospace Science and Industry Corporation,which is designed and implemented according to the actual needs of enterprises.It has a high-performance computing cloud-based job scheduling system with computing job submission.Computational job monitoring,computational job retrieval,virtual application remote invocation,user rights management,software license monitoring,cluster node monitoring and other major functions can meet the user's computing and management needs.The characteristics of the system platform are as follows: First,the job scheduling method combining multiple strategies is implemented,and the preemptive scheduling can be performed based on the priority of the job,and the reserved backfilling strategy is designed for the long wait situation of the operation,thereby improving the high performance.The actual efficiency of the computing system;secondly,based on the B/S architecture,the user can schedule the job through the web interface,and can graphically view the execution of the job in real time,so as to timely modify the calculation operation and make the user's operation more for the sake of simplicity.The main work of the thesis is as follows:1.Analyze the research status of high performance computing cloud jobscheduling system at home and abroad,analyze and compare eachscheduling strategy and choose;2.User requirements are analyzed and allover architecture is designed;3.The scheduling strategy,other main functions and database are designed;4.The main functions of the system are implemented;5.The summarize is given. |