Research On Reliability Enhanced Service Deployment In Mobile Edge Environment

As a distributed computing architecture,Mobile Edge Computing(MEC)migrates data processing and storage centers to edge nodes close to users and data sources,reducing core network bandwidth resource occupation and re-ducing service delays,greatly improving user service quality experience.At present,in mobile edge computing,container-based virtualization is usually used for service deployment to improve resource utilization.However in this deployment mode,if a physical server goes down,multiple containers deployed on it cannot provide services normally.To cope with this situation,this paper mainly study how to deploy services in the mobile edge environment,which can achieve the purpose of effectively improving the reliability of services un-der the premise of minimizing the number of physical machines.Due to the wide distribution of different edge node locations in the mobile edge environ-ment,communication and coordination between services in different locations need to be considered during deployment,so the deployment of services with enhanced reliability in the mobile edge environment faces many challenges.To deal with this problem,the contents of this paper are mainly divided into two parts.The first part addresses the problem of deployment service fail-ure caused by physical machine downtime in the mobile edge environment,and proposes a microservice-oriented container k fault-tolerant deployment method to ensure that the system provides services normally in the event of physical machine failure.The research modeles the problem of k fault-tolerant deploy-ment and multi-k fault-tolerant deployment as a combined optimization prob-lem firstly,and solves the problem by improved artificial bee colony algorithm.The final simulation experiment results show that the algorithm greatly im-proved the reliability of the service while controlling the use of physical re-sources.The second part studies the deployment method of the combined ser-vice with enhanced reliability.In a combined service,the communication band-width between different services cannot be ignored.Therefore,in addition to the deployment of containers,the bandwidth reservation on the link should also be considered during service deployment.This research first modeles the prob-lem as a cross-domain reliable virtual network mapping problem.At the same time,in order to enhance the reliability of the system,the original virtual request network is expanded to a survivable container infrastructure by additional re-sources to back up,and uses the k core decomposition plus packet mapping method at the same time,which makes service deployment more suitable for edge computing environments.Finally,a strategy-based reinforcement learning method DSVNM is used to solve the problem.Experimental simulation results show that this deployment strategy can provide better long-term revenue-cost ratio and higher resource utilization than conventional service deployment al-gorithms.