Intelligent Root Cause Analysis and Incident Diagnostics for Microsoft Fabric and Power BI through Telemetry Correlation and Kusto Query Language

Abstract Large-scale analytics platforms such as Microsoft Fabric and Power BI play a critical role in enterprise decision-making, making rapid and accurate incident diagnostics essential for maintaining reliability and user trust. However, incident investigation in these platforms remains challenging due to the volume, heterogeneity, and siloed nature of telemetry generated across data ingestion pipelines, semantic models, query execution engines, and shared capacity layers. Existing approaches rely heavily on manual analysis and ad hoc querying, resulting in prolonged diagnostic times and inconsistent root cause identification. This paper presents an intelligent root cause analysis and incident diagnostics framework that leverages multi-layer telemetry correlation and Kusto Query Language–driven analysis. The proposed approach unifies logs, metrics, traces, and platform-specific signals into a coherent analytical model and operationalizes reusable, dependency-aware KQL diagnostic templates to guide systematic investigation. An inference layer ranks probable root causes based on correlated telemetry patterns, enabling faster and more accurate diagnosis of analytics incidents. Experimental evaluation using controlled incident simulations demonstrates that the proposed framework significantly improves root cause identification accuracy and reduces mean time to diagnosis compared to traditional manual RCA workflows. Statistical analysis at a 95% confidence level confirms the reliability of these improvements. The results highlight the potential of intelligent, query-driven telemetry correlation to transform incident diagnostics in modern analytics platforms, offering a practical foundation for more reliable and scalable operations in Microsoft Fabric and Power BI-like environments.