| Strong military strength is a solid guarantee for national security.In peacetime,military exercises are usually used to improve the combat capability of troops.However,military exercises have high cost and cannot reproduce the combat plan.With the continuous development of simulation technology,people tend to conduct exercises through combat simulation.Compared with traditional exercises,it has the advantages of low cost,safety and repeatability.The combat environment can be simulated by constructing the combat simulation system,and can evaluate the battlefield in real time,compare the advantages of various combat schemes,improve the command ability of commanders.The combat simulation system is developing towards systematic and large-scale operation,and the number of members in the simulation system keeps increasing.At the same time,multiple simulation experiments are needed to improve the test efficiency of simulation samples.Under the above background,the current simulation resource management and control method has exposed the following problems: the characteristics of the simulation system require simulation resources to be transmitted and shared among multiple nodes,the current transmission mode of network and physical media requires manual intervention and consumes a lot of transmission time;During the deployment of simulation experiment environment,the experimenter needs to complete the environment dependence configuration and network configuration in multiple nodes,which leads to low deployment efficiency and the risk of configuration errors;In the running process of multiple simulation experiments,it is necessary to implement precise anomaly monitoring of multiple simulation environments.If there is no anomaly handling mechanism to ensure the sustainable operation of simulation experiments,a lot of experimental time and computing resources will be wasted.Therefore,an efficient simulation resource management and control method is needed to support the environment construction and operation control of simulation experiments.This paper uses the microservice architecture based on the characteristics of multi-member distributed deployment of combat simulation system,uses Open Stack cloud platform and Ceph distributed storage system according to the problems mentioned above.The main contents of the work are unified cloud management of combat simulation resources, automatic construction of multiple simulation system cluster experiment environment,anomaly monitoring and automatic handling of multiple simulation system cluster,mainly including the following aspects:Firstly,under the background of multiple simulation experiments,deeply analyze the problems in unified management of combat simulation resources,the construction and anomaly handling of multiple simulation system cluster experiment environments.Then use Open Stack cloud platform and Ceph distributed storage system,put forward the combination of static and dynamic control microservice architecture of simulation resources,dynamic microservices come online dynamically with the generation of multiple simulation nodes in the simulation system cluster.By using virtualization technology,the original control granularity extends from multiple simulation processes to multiple simulation nodes,then implement control over multiple simulation system cluster experiment environments.Secondly,in order to locate and acquire simulation resources automatically during the construction of multiple simulation system cluster experiment environments,classify the current combat simulation resources from the logical level and design the unified description method of simulation resources.Simulation resources can be quickly located in the cloud platform based on multiple groups of attributes.The simulation resources are transferred to Ceph distributed storage system as RBD block device from physical level,and the storage format of simulation resources is analyzed according to the criterion of resource acquisition time.The test results show that the time consumption of simulation resources acquisition can be effectively reduced.Thirdly,the simulation system cluster experiment environment build process is divided into a variety of service functions from the perspective of microservice,including the operations of the virtual machine level and process level.Rely on the overall control of static microservices and the fast response of dynamic microservices in each simulation system cluster,multiple simulation system cluster experiment environments can be automated build after requirements issued.A simulation system is taken as an example to test the construction of multiple simulation system cluster experiment environments,the feasibility and time efficiency of this method are verified.Lastly,for the purpose of anomaly monitoring and automatic handling of multiple simulation system clusters,rely on the dynamic microservice of each simulation node to achieve the accurate anomaly monitoring in multiple simulation system clusters.Based on the characteristics of synchronous simulation of multiple simulation members in the simulation system,use the Libvirt and RBD to achieve the snapshot and recovery of experimental state of simulation system cluster,and validate the sustainability of the simulation experiment after state recovery.Based on the above mentioned anomaly monitoring and cluster experiment state recovery of simulation system,the automatic anomaly handling mechanism can effectively guarantee the sustainable operation of the simulation experiment after anomaly occurs.It reduces the waste of experimental time and computing resources caused by anomaly. |
