A Novel Approach for Energy-Efficient Container Migration Using GNBO

Abstract

Cloud services are increasingly becoming available through containers because of their scalability, portability, and reliable deployment, particularly in microservices and smart vehicles. Due to the diversity of workloads and cloud resources, the scheduler component of cloud containers plays a crucial role in optimizing energy efficiency and minimizing costs. The growing demand for cloud services poses a challenge in terms of energy consumption. It is possible to optimize energy consumption on servers by utilizing live migration technology. This study aims to propose a hybrid model that will facilitate the migration of containers from one server to another using gradient descent Namib beetle optimization (GNBO) algorithms and thereby reduce the amount of energy consumed by cloud servers. The work is carried out by cloud simulation with physical machines (PMs), virtual machines (VMs), and containers. The tasks are allocated to VM in a round-robin manner. The actor-critic neural network (ACNN) is used to predict the load of PMs. Overloading and underloading conditions are determined based on the load. A hybrid optimization algorithm, GNBO, calculates the optimal solution based on predicted load, migration costs, resource utilization, energy consumption, and network bandwidth. This results in a load of 0.177 millions of instructions per second (MIPS), migration costs of 10.146 J, and energy consumption optimized to 0.068 W.