A Hybrid approach for containerized Microservices auto-scaling
Souheir Merkouche, Chafia Bouanaka · 2022
microservices are subject to an unpredictable and variant workload that challenges their quality of service. There-fore, several approaches emerge in the literature proposing auto-scaling policies as a promising solution to deal with such unpredictable increase/decrease of the workload. However, they generally deal with a specific and unique quality of service as cost-aware, latency-aware and other specific quality-awareness. We propose a multicriteria approach for microservice-based applications' auto-scaling. By means of a layered architecture, we present a hybrid auto-scaling solution that ensures an autonomous auto-scaling of microservices to enable them maintaining their specific required qualities. Moreover, a cooperative policy is applied to maintain the system overall qualities by identifying a compromise plan when the adaptation concerns different qualities of the system. The present work combines the efficiency of the hybrid strategy with the multicriteria selection principle to reach an autonomous-cooperative auto-scaling approach.