Resource-Aware Least Busy (RALB) Strategy for Load Balancing in Containerized Cloud Systems

Abstract

Containers are a key technology in modern cloud environments. They provide a lightweight manner to package and deploy service applications compared to virtual machines, with an isolated processing approach. The state-of-the-art containers LB algorithms do not take into account efficient mapping between container requirements and servers' capabilities, resulting in wastage of resources and increased response time, which negatively impacts the overall QoS. This manuscript proposes a new LB algorithm for containers called resource-aware least busy (RALB). RALB not only considers the current load of each node (as per the least connection) but also takes into account the current resource requirements of containers. It assigns containers to the least busy server with optimal capacity from a resource perspective. The authors implemented a prototype of the algorithm, and simulation results shows that more even distribution of workload and better utilization of resources across the nodes can be achieved, resulting in lower latency and high reliability compared to the randomized algorithm.