Adaptive Risk-Aware Energy and Integrity Optimization for Sustainable Cloud Resource Management

Abstract

Cloud data centres must reduce energy use without weakening the controls that protect data integrity. This paper presents an adaptive risk-aware resource-management framework that couples validation intensity, workload prediction uncertainty, VM consolidation, and migration decisions within a five-term objective covering energy, residual integrity exposure, SLA stress, validation overhead, and migration cost. The revised methodology defines system entities, state and decision variables, placement and capacity constraints, risk and uncertainty construction, normalization, candidate evaluation, and stability controls. The controller was evaluated in a reproducible 720-hour controlled simulation using 100 normalized hosts. To improve robustness beyond a single scenario, 30 independent stochastic runs were executed with identical paired inputs for all policies, and 95% confidence intervals, paired statistical tests, ablation experiments, and coefficient-sensitivity analysis were added. Across the 30 runs, the proposed adaptive policy used 101.34% relative energy, achieved an integrity-exposure index of 6.96, an SLA-violation rate of 13.55%, and validation overhead of 0.77%. Compared with the prior joint policy, it reduced integrity exposure by 18.4% and reduced SLA violations by 0.26 percentage points, while increasing relative energy by 0.32 percentage points. Compared with fixed validation, it reduced energy and validation overhead while accepting higher residual exposure. The results demonstrate a controllable and statistically repeatable trade-off rather than a single-objective optimum, and the paper provides an implementation mapping for AWS-style and orchestrated cloud environments.