Joint optimization of delay and cost for microservice composition in mobile edge computing

With the development of software technology, some complex mobile and Internet-of-Things (IoT) applications can be constituted by a set of microservices. At present, mobile edge computing (MEC) has been used for microservice provision to achieve faster response speed and less network pressure. Based on container technology, microservices can be easily deployed in the MEC environment, while multiple microservice instances in multiple locations need to be selected to provide services for a large number of users geographically distributed. How to fully consider the service response time, scheduling the startup and running strategies of microservice instances with the least resource cost for multiple mobile edge servers is the core problem of microservice composition. In this paper, we propose a multi-objective evolutionary approach (MSCMOE) based on improved NSGA-III to minimize the service access delay and network resource consumption in the process of microservice composition. In order to maintain the diversity of the population, we use the improved reference point strategy to enhance the computational efficiency of seeking elite solutions in the non-dominated layer. Experimental results based on a real data set of Shanghai Telecom demonstrate that MSCMOE can effectively reduce network resource consumption while reducing service request time.