Supercomputer System Environment And Running Policy Research And Application On DAWNING 4000A Supercomputer

With the rapid economic development of China, national strength has increased. As an important information-processing infrastructure, supercomputers expand the scope of its application. Cluster technology is developing rapidly, so it is more prevalent to be deployed on supercomputers. Super-cluster systems have become the basic research equipment in some large institutions, universities and research institutes. It plays a decisive role in the fields, which need intensive, massive data processing and engineering simulation.Efficiency and stability of Super-cluster computer depend on the system environment build by reliable file system, the overall synchronized user manage system and efficient job-scheduling software, and also depend on effective running policy with the characteristics of the actual business. They work together to manage all the computing resources for real-time monitoring, management and scheduling. All the resources can be the most efficiently used, and play all polymerized abilities.Although at present, there are many research papers about hardware and operating system of the super-cluster computer technology. But there is little actual application aspect research, especially the system environment and running policy research aims at the actual business. Therefore many cluster computer systems has only installed common operating system and job scheduling software. They can run some simply parallel program, but the actual efficiency is very low.This thesis takes the opportunity of the Dawning 4000A supercomputer system installation and operation to reach 10TB size of the supercomputer system. We research each side of system environment and running policy, such as file system, user system, job scheduling software, resources group, scheduling queues, information security and so on. All analysis...