Research Foundations for the Noir Chaos Engineering Platform: A Systematic Survey and Architectural Gap Analysis

Modern cloud-native applications are constructed as polyglot graphs of independently containerized services communicating overdynamic networks. Conventional unit and integration testing methodologies evaluate services under idealized runtime baselines, failing toexpose catastrophic failure modes that emerge under downstream latency, packet loss, or container resource starvation. This paper presentsa structured literature review of 22 foundational and recent publications(2015–2026) contextualizing the architecture of Noir, an autonomous,developer-first chaos engineering and resilience evaluation platform.Noir synthesizes static workspace profiling (Lynx), deterministic container perturbation via Linux traffic control and control groups (cgroups),external steady-state probing with strict recovery-time objective (RTO)validation, and algorithmic A–F resilience scoring with an explicit inconclusive statistical confidence gate. We map the literature across fivethematic dimensions, contrast Noir with existing state-of-the-art frameworks, and identify three crucial research opportunities unaddressedin prior work: empirical calibration of composite resilience gradingbands, formal sample-count statistical confidence gating, and statictopology-guided fault space pruning.