Multi-objective formulation of efficient microservice deployment in Kubernetes with dynamic resource allocation using hybrid nature-inspired algorithm

Abstract

Traditional Kubernetes microservice deployment methods often suffer from static resource allocation, leading to inefficient resource usage, high operational costs, and an inability to handle dynamic service dependencies in real-time. To address these challenges, this framework proposes a Hybridized Clouded leopard and Cock-hen-chicken Optimization (HCCO) model. The primary innovation of this research is the development of an adaptive switching framework that integrates Clouded Leopard Optimization (CLO) for robust global exploration with Cock-hen-chicken Optimization (CHCO) for hierarchical local refinement. The rationale for selecting this hybrid approach over other meta-heuristics, such as the Black Hole Algorithm, is its superior ability to avoid premature convergence in high-dimensional search spaces; while the Black Hole Algorithm often struggles with maintaining diversity during rapid resource shifts, HCCO utilizes a dynamic switching variable to ensure a balanced and faster search process. Experimental validation shows that the developed work achieves significant numerical improvements, including a throughput of 91.62%, a resource usage cost reduction of 9.61%, and a minimized execution time of 17.9 min. Ultimately, the HCCO framework provides a highly efficient and scalable solution for optimizing cost and performance in dynamic, real-time cloud environments.