Research On Microservice Requests Dispatching For Edge Computing Systems

Microservice architecture can split complex software services into fine-grained microservices,and develop,deploy,manage and update these microservices in a distributed manner,so as to improve the sustainability,agility,flexibility and scalability of software services.With the emergence of edge computing paradigm,these microservices can be deployed at edge servers,thereby reducing network transmission overhead and improving the quality of service experience.However,due to their limited computing and storage resources,edge servers can only be deployed with a limited number of microservice instances,moreover,service requests are distributed unevenly in edge computing systems,how to reasonably dispatch a group of microservice requests with invoking relationship contained in a microservice-based application to appropriate edge servers or cloud servers to reduce application response delay,and ultimately improve the quality of services has become a key problem to be solved urgently.In addition,because the traditional centralized optimization model relies too much on the central control node,it is easy to cause single point of failure and performance bottleneck problems.How to make full use of the distributed characteristics of edge computing and microservice architecture to solve the problem of microservice requests dispatching decision-making in a decentralized optimization mode has become the main research target of this paper.In order to solve the above problems,in this paper,we propose a game-theoretic mechanism for achieving efficient microservice requests dispatching for edge computing systems in view of the wide application of game theory in the design of lowcomplexity decentralized mechanisms.We first formulate the microservice requests dispatching problem among edge servers as a decentralized non-cooperative game model.In this model,application requests are no longer simply processed by one service,but are implemented by multiple microservice calls with associated dependencies.At the same time,channel communication between servers and between users and neighboring edge servers is considered.The queuing of service requests at servers is simulated,and the service requests are not processed in a simple mode of immediate processing,but need to go through the process of queuing and waiting to be executed.Then,after the construction of the non-cooperative game model is completed,this paper proposes the MRDG game algorithm,in which the edge server obtains the dispatching strategy of its respective microservice requests through multiple rounds of game and competition.When no edge server can unilaterally change its dispatching strategy to further improve its benefits,it indicates that the system has reached a Nash equilibrium state.In this state,all edge servers will obtain a mutually satisfactory microservice requests dispatchng decision.In the process of the game,this paper comprehensively considers the dispatching and optimization of application requests in the serving overlapping areas of edge servers,and proposes the OROA optimization algorithm to further reduce the response delay.Finally,we conduct a series of experiments to evaluate the performance of our approach.The result shows that our approach can effectively reduce the average response time of microservice-based applications.