Availability-Constrained Application Deployment in Hybrid Cloud-Edge Collaborative Environment

Cloud computing offers strong availability and lower cost, while edge computing has lower delay. Deployment of applications by placing microservices in containers in a cloud-edge collaborative environment is adopted by more and more enterprise application providers. For users, they care more about application response time and application availability. For application providers, they also need to save deployment costs to the maximum extent. Therefore, the application deployment in hybrid cloud-edge collaborative environment is a multi-objective optimization problem. In this paper, a genetic algorithm named DP-GA based on improved NSGA-II has been proposed to solve the multi-objective NP-hard problem. We balance the two objectives of minimizing deployment cost and average response time under availability constraints. Using the real dataset of Shanghai Telecom, the experimental results show that the proposed DP-GA is superior to the existing methods, reducing average response time by about 35% and saving deployment cost by about 15%.