REACTOR: Reliability Engineering with Automated Causal Tracking and Observability Reasoning

Reliability engineering aims to ensure that systems perform as expected over time, yet it faces various challenges in identifying and mitigating potential failures. We introduce REACTOR, an advanced framework prioritizing automated causal tracking and observability reasoning to improve reliability analysis. REACTOR uniquely utilizes a dual-layer architecture to facilitate the identification of failure sources through thorough causal analysis and subsequently assesses the ramifications of these failures on system performance through observability reasoning. This framework minimizes reliance on manual interventions, enabling users to achieve a deeper understanding of the reliability of complex systems. We employ sophisticated machine learning techniques to bolster the detection of anomalies and pinpoint their root causes, fostering a proactive approach to reliability management.