Instrumenting at 10 years per second
summary
Claims a concrete pipeline: KSI (config accelerated custom instrumentation) turns declarative capture configs into runtime hooks, using Python metaclass wrapping to inject capture logic without editing app code. Pilot moves one microservice from 1% to 60% instrumentation in under a week, surfaces a premium-feature leak fixed in 15 minutes, and replaces a 3-month-per-service estimate that summed to 10 years for 40 services with a scalable path. Tools mentioned include DataLog/DD trace style telemetry and trace/log/error integration.
AI-written summary · about
chapters (7)
- 0:00Introduction to Accelerating Manual Instrumentation
- 1:37The Power of Manual Instrumentation: A Story
- 5:03Identifying the Problem and Initial Solution
- 12:19Developing a New Approach: Component Interaction and Data Exchange
- 15:34Implementing the Solution: Meta Classes and Configuration
- 21:53Results and Future Plans
- 26:20Conclusion and Contact Information
related talks
01h 00m“Source Instrumentation for Monitoring C++ in Production”
26mA Year of Observability Across the Toolstack: Breakthroughs, Challenges, Improveme...
32mDistributed Tracing in Action: Our Journey with OpenTelemetry
46mObservability in the real world: Improve systems w/iterative approach
20mHow to Think About Instrumentation Overhead
04mSponsored Keynote: Zero-Code Observability: Close the Coverage Gaps That Cause Outa...
46mPintrace: A Distributed Tracing Pipeline
28m"What Happened in Production?! Instrumenting with OpenTelemetry"