Design And Implementation Of Serverless Based Edge Computing Platform

As edge computing moves to practice,edge computing platforms have also flourished.Edge computing platforms provide a set of edge service related components,effectively solving the edge service deployment,operation and maintenance.And finally provides a reliable execution environment for edge services.But at the same time,the current mainstream virtual machine computing platforms and container computing platforms require services stay in the background to monitor user requests,which consumes a lot of resources.The scalability of edge computing services also requires a lot of manual operation and maintenance,which introduce a lot of troubles in the edge computing environment.In general,these mainstream computing platforms have troubles in resource utilization and convenient maintenance in the edge computing environment.The Serverless computing paradigm can allocate resources on demand and expanding computing instances automatically,it could solve the problems of edge computing.The Serverless computing paradigm has a great application prospects in the edge computing feild.The mainstream Serverless computing platform is designed for cloud computing centers.Due to the difference between the edge computing environment and the cloud computing environment,directly application of the Serverless computing paradigm in the edge computing environment will encounter various problems,such as the fine-grained function model will lead to delay amplification,warm-up model can also causes waste of resources,service migration does not be supported.Therefore,the application of the Serverless computing paradigm in the edge computing environment requires a redesign of the computing platform.Based on the various bottlenecks of the current computing platform and the huge potential of the Serverless paradigm in edge computing,this thesis proposed and implemented the service function graph model and service warm-up model,which effectively solves the problems encountered by Serverless in the edge computing environment.The service function graph model optimizes the delay and resource utilization of the Serverless computing paradigm in the edge computing environment,and further supports service snapshots on the basis of the service function graph.The function warm-up model proposed in this thesis enhance the traditional Serverless warm-up model.It divides functions into delay-sensitive functions and non-delay-sensitive functions,and performs warn-up only on delay-sensitive functions.This simple warm-up model can effectively saves the resource loss.Based on targeted improvements to Serverless technology,this thesis solved the problem of application Serverless in the edge computing field,and then completed the design and implementation of edge computing platform system,which meets the needs of edge computing environment.