MDSGA: A Limited Resources Deployment Strategy of Microservice-based Application based on Genetic Algorithm

In recent years, the deployment strategy of microservice instances has become one of the important issues concerned by microservice applications. Most solutions assume that resources are sufficient to satisfy all current user requests, and the total response time of user requests is a constant. When user requests cannot be executed concurrently due to limited resources, the time to complete user requests is a variable and needs to be considered in the deployment strategy. This paper proposes a limited resources microservice deployment strategy based on improved genetic algorithm. By improving the individual coding method, each individual has a microservice deployment substring and a user request scheduling substring, so that the strategy can calculate the optimal deployment strategy based on the deployment distribution status and scheduling relationship. This paper also proposes a calculation method for user request response time considering the microservice chain under limited resources, and calculates the user request response time considering the blocking state time of microservices. Experimental results show that the strategy can effectively calculate the microservice deployment scheme with low total response time of user requests, low load balancing rate and high resource utilization rate under the condition of limited resources.