The Role of Context in Observability
summary
Shows how metadata actually alters ops decisions: raw Prometheus text exposition labels and a StatsD gauge reveal queryable fields and metric types; diurnal periodicity (basement temperature, gaming concurrent users differing by weekday/weekend) changes alert thresholds; enrichment via external lookup adds service_value to checkout metrics; feature extraction pulls host, service, cpu→compute type; NLP clustering groups noisy messages by stem; a Raspberry Pi experiment shows third‑party change data can surface CICD-caused incidents.
AI-written summary · about
chapters (6)
related talks
29mObservability; a gentle introduction
36mMetadata: The Secret Sauce for Full Observability
40mDemystifying Observability and Monitoring
48mMetrics All the Way: Data Driven Operations
41mMetrics Applied
35mObservability: Building Confidence & Answering the What, Where, How
27mObservability with Prometheus and beyond
52mObservability: it's not just an ops thing