Watch a 48-minute conference talk from QCon exploring how a development team experimented with building a Go-based service using only event tracing for observability, completely foregoing traditional logging. Learn about the journey of implementing Honeycomb for telemetry, the challenges and benefits of event-driven instrumentation, and the practical implications of this unconventional approach. Discover insights into error capture methods, unified telemetry patterns, metrics collection, and developer experience when working with traces instead of logs. Gain valuable perspectives on modern observability practices and how they can enhance system behavior understanding in production environments.
Building Observable Services with Event Tracing - A No-Logs Approach
Overview
Syllabus
Introduction
Setting the scene
The team
Bike shedding
Background knowledge
Experiments
Event First
New Words
Error Capture
Medium Cardinality
Cost
Patterns
Traces
No Logs
Metrics
Unified Telemetry
Conclusion
Questions
Practicality
Telemetry
Developer Experience
correlating logs and traces
sending logs to a second location
Unified Telemetry approach
Taught by
InfoQ
