Context and Correlation in Observability Through Graph-Powered Insights
summary
Splunk Observability derives an entity graph from OpenTelemetry signals (metrics from Kubernetes, traces for APM), persists relationships in Neo4j, and uses static and dynamic relations (one-to-many, many-to-many) to correlate across domains. It feeds entities to LLMs for context-rich root-cause analysis, implements an entity health grouping to reduce alert noise, and balances right-path vs read-path derivation tradeoffs; solution targets real-world scale (hundreds of thousands of pods) and aims to lower mean time to recover.
AI-written summary · about
chapters (8)
- 0:00Introduction to graph-powered observability
- 1:14Splunk observability platform overview
- 4:52Entity modeling and graph creation
- 7:01Entity health and alert grouping
- 8:44Troubleshooting with context and correlation
- 13:25Entity model examples and relationships
- 17:48Dynamic relationships and entity derivation
- 19:55Q&A and feedback
related talks
26mWhy Large-Scale Observability Needs Graph
52mGraphQL Observability with Elastic and OpenTelemetry
42mOpen Source for Better Observability
42mObservability can help you understand change
24mWhat Observability Can Learn From BI
12mOpen-source observability is the way forward, and here’s why
22mUnified Observability
08mMastering Observability: The Coralogix Maturity Blueprint