Research on Microservice Migration Mechanism for Mobile Edge Network

Microservice is a new type of service architecture different from monomer service architecture, which can improve the service quality of nearby users. However, the movement of users will lead to frequent switching of edge servers that provide services to users, increasing service delay and service cost. It is difficult to meet both the low latency requirements of users and the low cost requirements of service providers. Therefore, this paper aims to solve how to formulate a microservice deployment and migration strategy considering service response time and service cost, construct a microservice deployment migration model, abstract the problem into a Markov decision process, and use deep reinforcement learning algorithm to solve it. Finally, simulation experiments are carried out to compare the effect of deep reinforcement learning algorithm with that of heuristic algorithm and random selection algorithm, and to verify the performance of reinforcement learning algorithm in solving this problem.