A Group Genetic Algorithm for Resource Allocation in Container-Based Clouds

Containers have gain popularity because they support fast development and deployment of cloud-native software such as micro-services and server-less applications. Additionally, containers have low overhead, hence they save resources in cloud data centers. However, the difficulty of the Resource Allocation in Container-based clouds (RAC) is far beyond Virtual Machine (VM)-based clouds. The allocation task selects heterogeneous VMs to host containers and consolidate VMs to Physical Machines (PMs) simultaneously. Due to the high complexity, existing approaches use simple rule-based heuristics and meta-heuristics to solve the RAC problem. They either prone to stuck at local optima or have inherent defects in their indirect representations. To address these issues, we propose a novel group genetic algorithm (GGA) with a direct representation and problem-specific operators. This design has shown significantly better performance than the state-of-the-art algorithms in a wide range of test datasets.