The Research And Realization Of Resources Management And Task Scheduling Sub-System In Couple Distribute System

Distributed simulation system is the comprehensive infrastructure of hardware and software to provide reliable, consistent, transparent and extensible high-end computing power, it integrates with geographically distributed, heterogeneous and dynamic of the various high-performance computing resources, making the customers can use the system resources transparently, seamlessly and efficiently but not restricted by geographical boundaries, thus it can solve problems of kinds of Multi-machine load coordinate simulation. Assignments generally refers to a single task or parallel tasks, but coupling distributed task simulation is to put forward a new simulation tasks demand meeting the high-speed train digital simulation requirements, namely, the simulation of a task depends on one or more other tasks and there is Input/output coupling relationship between tasks and tasks. Undoubtedly, since it is distributed simulation, we need to build a suitable distributed simulation framework according to the characteristics of the simulation, to solve the management of the resources and working conditions as well as the simulation task scheduling during the process of simulation.This thesis regards high-speed train digital simulation platform as the research background. On the basis of analyzing user needs, the Resource management and Task scheduling sub-system based on the coupling of distributed simulation are designed and implemented. This sub-system mainly completes resource management, working condition management, simulation working condition and task scheduling of distributed system, what’s more,the failure of computing resources and the process of the working condition simulation running are dealt with, the failure maybe happened of the scheduling system itself is also recovered in this paper. Multi Agent technology is introduced in terms of resource management, by running agent software on a computing resource node to realize resources registration, cancellation, updating resources, resources recycling and resource fault detection etc; Management of Multiple working conditions at the same time are realized, including the condition priority sorting, conditions resource request message parsing, condition query, control, and the management of the working state etc; The working condition scheduling realizes conditions of pre-dispatch, couplers load balance and task scheduling, and provides external management interface of the scheduling strategy, for users to design their own scheduling strategy; Through the combination of agency, the scheduler can test drops in the system resources and some fault condition according to the working condition of simulation information offered by other sub-systems,It can restore the running working condition of system resources information and the working information according to the information stored in the database and file correctly restore after the system itself is turned off to restart, thus it can ensure normal operation of the simulation system. The file data used in the system, simulation data used in the process of the simulation and the computer resource information in the system are all stored by data base.It improves the stability of the system, and facilitates the management of information data. Scheduling system and other sub-systems can implement the network data communication between each other by UDP communication and user-defined communications protocol.This thesis firstly introduces the research background,significance, domestic and foreign research present situation and the main research work, Then, the demand analysis, the system’s design goal and function demand are described in detail; On this basis,each function module of this system are briefly introduced and the design and implementation of those function modules are also introduced in detail. Finally, a functional test of the whole system was made, and at the same time the whole work is summarized and prospected, the deficiencies and the next step research work are analyzed and pointed out as well.