Research And Implementation Of Boundless Application System Based On Container Technology

Limitations like network latency and expensive cost are driving cloud computing to the edge.Edge computing is an extension of cloud computing.Edge computing integrates computing nodes on the edge to cut the distance between computing and terminal data,which solves the problem that tradition cloud computing paradigm has strong dependence on the network.With the development of Internet of Things,the amount of edge devices is increasing as well as the computing resource of edge devices.Therefore,the scope of edge computing is gradually expanding as well.Many edge computing paradigms failed due to changes and problems caused by the expansion of the edge computing system.This paper proposes an edge computing architecture based on container technology called the Boundless Application System(BRO).BRO extends the two-layer architecture of the original edge computing to a three-tier architecture which aims to extend the boundary of edge computing,accommodate more edge nodes,enhance the computing power of the edge computing architecture,and bring the calculation closer to the terminal.BRO solves the following problems:1.BRO is implemented based on container technology and container orchestration technology.BRO applied container technology as a virtualization plan to solve the isolation problem of the underlying operating system of heterogeneous edge computing nodes.BRO optimized container orchestration technology to achieve precise orchestration of application services and resources in the edge cloud.BRO solved the problem of resource orchestration in complex edge environments.2.BRO proposes Dynamic Clustering on Edge(DCE)algorithm.It enables the edge nodes to be dynamically aggregated into edge clouds,and the edge layers in the edge computin g architecture are divided into smaller,more manageable multiple edge clouds.It solves the problem that applications,resources,and nodes are difficult to manage effectively as the amount edge nodes increase.The DCE algorithm solves the problem of dynamic clustering of movable and scattered edge nodes by guiding the autonomous clustering of edge nodes.In addition,the DCE algorithm also supports the elastic scaling of the edge cloud,which solves the problem that the edge cloud is of different scale and resources are not balanced generally.3.BRO proposes the Container Orchestration on Edge(COE)algorithm.The paper designed and implemented the edge node price model,the resource perception model,the optimization of the scheduling occurrence judgment mechanism based on load prediction and the multi-level scheduling mechanism.COE provides a precise,smooth,and efficient solution for container scheduling in edge clouds.BRO sloved the problem of unbalanced edge cloud scheduling,improved the scheduling accuracy,reduced the occurrence rate of invalid scheduling in the edge cloud,and improved the average resource utilization of the edge cloud.At the end of the paper,several experiments were designed based on an edge computing environment simulated using containers and virtual machines.The paper compared DCE algorithm with the maximal group-based edge-cloud aggregation algorithm,and verify the DCE algorithm in the edge computing background by comparing the convergence degree,local maximum delay and resource balance degree of the edge cloud aggregated by applying the algorithm.The paper compared COE algorithm with the container scheduling algorithm of Kubernetes in the background of edge computing and verified the DCE algorithm by comparing the efficiency,accuracy and average resource utilization of the edge cloud.The paper compared BRO with Kubernetes and verify the performance of the BRO by the average completion time of the task.