How AIOps Platforms Work

#72 - Managing SRE Toils Using AIOps and NoOps - Amrith Raj

Topic 7 of 9

Published Jan 17, 2022 · Clip 34:27–44:17 (9:50)

Amrith explains how AIOps platforms work: collecting metrics, logs, and traces; understanding full topology from front-end to node level; automatically baselining behavior so only true anomalies alert (not a routine Monday 9am CPU rise); condensing noisy events into meaningful alerts with user impact; and providing root cause analysis through causation rather than correlation. He illustrates with distributed tracing as tracking a person through CCTV in a skyscraper and the cat/pothole analogy for why correlation is misleading. He then details how Dynatrace's OneAgent automatically detects environments, technologies, and topologies, and references the Google authentication outage caused by storage quota going to zero.