Towards Net-Zero clouds: An SLA-aware, carbon-optimized, and hybrid resource management technique

Abstract

Cloud Data Centres (CDCs) constitute the foundation of digital services, but they are also a major source of carbon emissions and energy demand worldwide. In order to achieve net-zero infrastructures, environmental sustainability, energy efficiency, and service quality, in terms of Service Level Agreements (SLAs), assurances must be balanced. In this study, we present a new framework for hybrid cloud–edge resource management that is energy efficient, carbon-optimized, and SLA-aware. The three modules that make up our framework are: (i) a predictive migration engine that uses deep recurrent models to forecast workloads and place services proactively; (ii) a carbon-aware scheduler that uses forecasted and real-time grid carbon intensity to reduce emissions through placement; and (iii) a hybrid cloud–edge orchestrator that dynamically allocates workloads among edge nodes and CDCs according to sustainability objectives, compute intensity, and service latency. Comprehensive experiments on real-world traces shows that the proposed approach can considerably exceed state-of-the-art baselines, achieving up to 23% energy savings, 31% carbon reduction, and > 95% SLA compliance. By combining hybrid orchestration, carbon-aware scheduling, and predictive intelligence into a single comprehensive architecture, our framework advances the field of Net-Zero cloud infrastructures.