Risk-aware resilient cloud orchestration under correlated faults using adaptive dual-regime search with self-healing control

Key findings from this study

This research indicates that:

• Conditional value-at-risk penalisation in multi-objective orchestration successfully balances fault tolerance and latency constraints under correlated hardware failures.
• Event-triggered self-healing policies with bounded-switching guarantees reduce repair cost volatility without excessive controller switching.
• Statistically validated performance gains emerge across five stress scenarios combining correlated faults with workload surges, relative to established cloud orchestration baselines.

Overview

The paper presents a fuzzy hybrid reptile–mamba optimisation framework addressing resource allocation in cloud platforms subject to correlated hardware faults and competing performance demands. The framework formulates orchestration as a multi-objective risk-constrained optimisation problem integrating fault tolerance, response latency, and recovery overhead. A novel black mamba operator, derived from predatory strike kinematics, combines with reptile search under interval type-2 fuzzy control to balance exploration and intensification.

Methods and approach

The framework employs conditional value-at-risk penalisation to model risk across multiple competing objectives. A black mamba operator provides local intensification, fused with reptile search for exploration under interval type-2 fuzzy mode selection. An event-triggered self-healing policy incorporates a local sufficient-decrease guarantee for repair costs and hysteresis-based bounded switching for controller selection. Experimental validation used CloudSim Plus simulation with Google Cluster Trace data across 30 independent runs per scenario.

Results

Experiments across five stress scenarios—including correlated faults and bursty workloads—demonstrated statistically significant improvements in the proposed framework relative to baselines. Friedman and Wilcoxon nonparametric tests confirmed performance gains across resource allocation, latency minimisation, and recovery overhead reduction. The framework maintained feasibility under combined fault and workload stresses that induced constraint violations in comparative methods.

Implications

The risk-aware formulation via conditional value-at-risk extends cloud orchestration beyond nominal conditions, addressing practical scenarios involving simultaneous hardware failures and load surges. The integration of self-healing policies with formal guarantees on repair cost decrease and controller stability reduces manual intervention in degraded operational states. The approach establishes a pathway for incorporating bio-inspired operators with rigorous control-theoretic guarantees in cloud resource management.

Scope and limitations

This summary is based on the study abstract and available metadata. It does not include a full analysis of the complete paper, supplementary materials, or underlying datasets unless explicitly stated. Findings should be interpreted in the context of the original publication.