AIOps on Observability and Recent Trends: A Literature Review
AIOps ("Artificial Intelligence for IT Operations") is the application of machine learning and, increasingly, large language models (LLMs) to observability telemetry — metrics, logs, traces, and events — in order to detect anomalies, reduce alert noise, correlate signals, and localize root causes faster than manual operation sallow. This note surveys the principal method families for anomaly detection, clustering and correlation, and root-cause analysis (RCA), then traces developments from 2021–2025:transformer and foundation models for time series, LLM-based log/telemetry analysis,eBPF-native instrumentation, and the consolidation of OpenTelemetry as the vendor-neutral standard. Each substantive technical claim is tied to a primary source (arXiv ID/DOI orofficial docs). Where a claim could not be verified against a primary source, it is flagged explicitly in the text.