Research On Containerization Resource Management Technology Of Embedded System For Light Cloud

Due to the rise of the Internet of things(IoT)and the Internet of vehicles(IoV),the change of computing scenarios brings new opportunities and challenges to traditional cloud computing.Light cloud computing based on embedded devices is an effective solution to the computing requirements of the IoT and the IoV.On the premise of meeting the key performance indicators,how to design a unified resource management architecture for light cloud and how to effectively orchestrate and deploy light cloud resources are the key issues to realize the development and maturity of the IoT and the IoV.This paper focuses on the above issues and conducts research.The main work and innovations are as follows:Firstly,in the resource-limited scenarios of the IoT and the IoV,a distributed containerization resource management architecture based on embedded system is designed.The computing or storage services for the IoT and the IoV scenarios are referred to as light cloud computing.Embedded devices have the characteristics of dexterity,reliability and low power consumption.In addition,the container technology has the advantages of light,safe,fast and large-scale deployment.Therefore,the container resource management technology based on embedded system is an effective and convenient resource management scheme in the light cloud scenario.Secondly,combined with containerization,container orchestration,network function virtualization(NFV)and other technologies,the distributed containerization resource management experimental platform based on embedded system is designed and built,and the experimental analysis is completed.The experimental results show that the platform can provide the basic computing resources required by light cloud computing on demand.Finally,referring to the current popular deep learning algorithm,aiming at the timeliness of resource scheduling,this paper proposes a resource management scheme based on two different traffic prediction algorithms,and designs two different deep-learning training models,which can predict the traffic based on the historical traffic data and realize resources deployment in advance.The experimental results of the original resource management mechanism HPA and the LSTM-HPA and NN-HPA mechanisms based on the long short-term memory network(LSTM)and back propagation neural network(BPNN)algorithms respectively show that the proposed scheme can effectively solve the problem of resource load imbalance,and the final effect meets the service level agreement(SLA)requirements of the IoV.