Research On Microservice Scheduling Algorithm Based On Edge Computing

In recent years,with the continuous development of mobile edge computing,there have been many emerging and diverse edge applications(such as Internet of Vehicles,augmented/virtual reality games and lightweight deep learning tasks),which are extremely sensitive to latency High,requires a large number of network connections and lower latency,which will lead to the increasing size of data and business requests processed by the core network in unit time,challenging the computing power of mobile devices under limited resources.Because the data generated by these delay-sensitive applications must pass through the core network in a short period of time,the peak load of service access will put a lot of pressure on the edge cloud network load.The traditional mobile edge server has been unable to provide timely and effective Service.In order to meet the high requirements for delay-sensitive application services,on the one hand,it is necessary to reduce the energy consumption of edge-end devices,and dispatch modules with high computational volume to remote servers for execution;on the other hand,the computing modules must be as close to the data source as possible.To reduce the delay caused by data transmission between modules.Therefore,it is necessary to rationally optimize and expand edge cloud computing resources close to physical entities and data sources,enhance the core capabilities of service networks,computing,storage,and applications,and marginalize and localize new computing resources and cache resources.First,on the basis of existing service scheduling research,this paper proposes a new model of edge application services.Each application service is composed of many loosely coupled microservices.Application services can be represented by a weighted graph.The nodes in represent the microservices that can be scheduled,the edges in the graph represent the communication and transmission between the microservices,and the weights on the graph represent the computational cost and communication cost of the microservices executing in different locations,respectively.By analyzing the edge computing scheduling scenario,an application partition algorithm based on this model is proposed to minimize the total time cost of the system.Then,for the application scenarios of UAV-assisted edge computing,the partitioning algorithm is further extended,and a MCMO(min-cut matching offloading)scheduling algorithm is proposed,which forms multiple pairs on the side of schedulable microservices and computing resources such as drones.A selection model that allows microservices to match the choices made by computing resources such as drones.After extending the CloudSim cloud simulation program,the proposed model and algorithm are analyzed experimentally,and compared with other four scheduling methods,it is verified that the proposed algorithm can effectively use the computing resources in the system and reduce the completion of application services time.Finally,a lightweight edge computing platform was developed based on kubernetes,and the kubernetes scheduler module was designed and extended according to the models and algorithms proposed in this paper.The face recognition program was experimentally verified and demonstrated that the theoretical method was deployed in practical applications.Full applicability.