| （英） |
Microservices architecture is well-suited for agile development practices due to its high degree of flexibility and scalability, and this architecture involves the connection of small, independent services to enable the development of large-scale software. In the context of microservices architecture, each individual service is responsible for executing various processes. To further reduce latency and the amount of traffic, currently, it is considered to be used in an In-Network Computing (INC) environment where several types of computing process is performed by network devices. When deploying microservices in the INC environment, multiple microservices are used as a microservice chain through service chaining techniques. In this paper, we propose an optimal configuration for microservices that minimizes operational costs in the INC environment. The proposed optimal configuration formulates a cost-minimizing optimization problem with three latency constraints for the the entire microservice chain includes multiple microservices, each microservice, and the distributed servers. Then, by deriving solutions to the optimization problem, appropriate microservices are configured in the INC environment. In numerical examples, the performance of the proposed method based on genetic algorithm is compared with several methods to evaluate its effectiveness. |