Service-oriented Cloud Simulation Platform Resource Management

At present,the traditional stand-alone simulation tools lack a unified resource man agement platform,and the barriers to entry are relatively high.The massive data gene rated during the simulation process,such as the intermediate data calculated during the simulation process,has not been centrally and effectively managed.Data is a very va luable reference material for product development.This article combines cloud computi ng to encapsulate simulation design,simulation calculation,and data management,ope n up the data interfaces of the computing layer,storage layer,and scheduling layer,in tegrate existing hardware and software resources,provide a unified resource manageme nt and scheduling mechanism,and achieve design experience.Standardized preservation and unified management of resources.1.Cloud simulation platform architecture design.The architecture design of clou d-based,open and shared network collaborative simulation service platform is realized,which supports multiple users to use the platform for simulation modeling and demo nstration at the same time.2.Cloud simulation platform data resource management.In view of the fact that the cloud simulation platform has severe competition for read and write locks under h igh concurrency,which has led to the problem of reduced service quality of the cloud simulation platform,the cloud simulation platform lock mechanism has been optimize d to reduce the probability of lock competition,and the cloud platform has been impr oved The reading process realizes the network reading,short-circuit reading and file z ero-copy reading layered reading,which improves the data reading efficiency.3.Cloud simulation platform computing resource management and scheduling.Ba sed on the in-depth study and analysis of the current mainstream scheduling mechanism,improvements and optimizations have been made.After testing,the improved sched uling mechanism can make the computing resources of the cloud simulation platform more evenly distributed among the cluster hosts.At the same time,in order to solve t he lack of stability of the cross-border point calculation of the cloud simulation platfo rm,cross-node calculation based on HDFS is realized.Based on the above work,completed the work of service-oriented cloud simulation platform.This paper implements the design work of the simulation cloud platform ba sed on the open source big data framework Hadoop and the resource scheduling fram ework Yarn.At present,the entire simulation system is running well,which greatly su pports the development of a variety of new aircraft models and has important national defense construction significance.