A Microservice-based SaaS Deployment in a Data Center Considering Computational Server and Network Energy Consumption

In a data center, deploying a microservice-based SaaS will result in a spike in energy usage, since the computation servers where the microservices are deployed will consume more energy as a result of an increase in computation workload. The data center's communications network will also consume more energy due to the increased communication among the microservices of the SaaS. Due to the fact that microservice-based SaaS deployment is handled by the developer of the microservice-based SaaS and can only be deployed on virtual machines rented by the developer, this issue cannot be taken into account in traditional microservice-based SaaS deployment methods. In this paper, a new microservice-based SaaS deployment method is proposed. The new approach relies on the data center determining where microservices should be deployed. The new microservice-based SaaS deployment method considers the energy increase in the computation servers and in the communication network. The microservice-based SaaS deployment problem is a combinatorial optimization problem. Thus, a genetic algorithm with repairing mechanism is proposed to solve the problem. Compared to traditional deployment approach, the proposed method is capable of reducing the energy consumption associated with microservice-based SaaS deployment by 37.55%.