| The upgrading of information-based operations has put forward higher requirements for military cloud platforms.There are two main problems in the operation of the current cloud computing technology for the battlefield analysis and system（that is,the battlefield analysis application in cloud platform）: The first is that resource allocation scheme is inflexible and the resource utilization rate is low;The second is the lack of data security,and the threat of interruption to applications.Therefore,this paper based on K8s（Kubernetes）container cluster virtualization technology and Fabric blockchain technology,develop K8 s cloud platform based on battlefield analysis applications.The platform can effectively solve the above problems,realize the effective utilization of resources,continuous operation of applications and reliable storage of data.This paper designs a K8 s cloud platform suitable for battlefield analysis applications.The platform achieves high availability by improving resource scheduling strategies and K8s-oriented container application migration;Realize high-trust storage of key data through K8s-based Fabric applications;Uses Golang language for secondary development of platform functions to achieve visual management.The main research contents are as follows:Firstly,on the basis of analyzing and running the default resource scheduling policy of K8 s container cluster,In view of the resource requirements and characteristics of battlefield analysis applications,this paper improves the scheduling strategy from three aspects: Pod priority,GPU resources,and video memory resources;and comprehensively considers the degree of matching between application demand and node resource,So as to realize the full utilization of node resources in the cloud platform and the load balancing of the cluster.Secondly,the online migration method of virtual machine and the container migration method between single hosts are studied.The migration method of container application for cluster is optimized,and the file system is compared and migrated by using the layered characteristics of docker to reduce the transmission of redundant data;Optimize the conditions of memory downtime migration to avoid the problem of unlimited iteration.Then,the underlying technology of Fabric blockchain is studied,and the K8s-based blockchain application is implemented.So that the basic components can be started directly with the network,and the scalable Chaincode can be started on demand through the control components,which further completes the trusted storage of key data.Finally,Golang language is used to realize the function development of cloud platform,including four modules: platform overview,applications management,images management and data trust management,providing users with a visual operation interface for easy use.After the functional test and performance analysis of the platform,the K8 s cloud platform based on the battlefield analysis application designed in this paper has perfect functions and runs stably.The improved scheduler improves the load balancing efficiency of individual nodes and clusters;Container transfer in the amount of data transferred,transfer time and downtime has improved the performance of the three aspects.The platform can meet the actual demand in terms of function and performance,and has achieved certain application innovation in battlefield analysis and simulation scenarios,which has good application value. |
PDF Full Text Request