Event Correlation: A Fresh Approach towards Reducing MTTR
summary
Describes LinkedIn’s Event Correlation Engine that algorithmically identifies the service responsible for incidents in a large microservices environment. It analyzes downstream latencies, error metrics and monitoring events across caller–callee relationships, and uses dynamic thresholds learned from the last 30 days of data to recommend the right service and reduce MTTR and escalations.
AI-written summary · about
chapters (14)
- 0:00Start
- 1:30Introductions
- 2:37Problem Statement
- 7:26Architecture considerations
- 12:11Microservices knowledge
- 13:52How do we map service
- 17:54Approaches that we tried
- 22:13K-Means: How it works
- 24:40Predict score
- 26:18Typical Workflow
- 34:36Recommendation Engine
- 36:11Event Correlation System
- 38:43The Big Picture
- 41:36Key Takeaways
related talks
30mReducing MTTR and False Escalations: Event...
37mWeathering the Storm: How Early Warnings Save the Farm
20mSpike Detection in Alert Correlation
42mMachine Learning In Observability Systems: From Insights To Action
33mThe Problem with MTTR: Learning from Incident Reports
30mComposing Events in a Distributed Environment
15mIntelligent Incident Management in DevOps
26mML Monitoring & Observability