Estratégia de alocação dinâmica de recursos no Kubernetes para ambientes multi-inquilinos

Abstract

The growing adoption of cloud-native applications has intensified the demand for efficient and secure multi-tenancy solutions in Kubernetes environments. However, dynamic resource allocation in shared environments presents significant challenges, requiring a balance between maximizing resource utilization and ensuring strict isolation between tenants. This study addresses these challenges by analyzing multi-tenant usage in Kubernetes, identifying its potential and limitations. Based on this analysis, we propose the Multi-Tenant Strategy for Kubernetes (EMK), which incorporates a Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS)-based algorithm for multi-criteria decision-making in dynamic resource allocation. This approach is integrated into a Kubernetes operator, enabling more efficient and automated tenant management while optimizing resource distribution in multi-tenant scenarios. The experimental evaluation was conducted in a Kubernetes cluster environment, considering different workload scenarios (light, moderate, and heavy). Performance indicators such as response time, allocation success rate, cluster flexibility, and adaptability were analyzed, all collected through Kubernetes’ native monitoring tools. The results demonstrated that the EMK reduces average latency and improves system stability under varying load conditions when compared to traditional allocation, highlighting greater efficiency and resilience in multi-tenant resource management.