Observability 2.0: Transforming Logging & Metrics • Charity Majors & James Lewis • GOTO 2024

Observability 2.0 focuses on wide structured events with preserved context, moving beyond traditional metrics, logs, and traces

Modern debugging requires understanding system complexity outside the code, not just within it - the traditional debugger is now “the debugger of last resort”

High cardinality and business metrics need to be blended with application/system metrics for effective observability

Cost management in observability is becoming critical - traditional approaches using multiple tools and inappropriate data models are unsustainably expensive

Canonical logging provides a single source of truth, collecting all context in one wide structured event rather than scattered across multiple logs

Engineering culture needs to shift from only looking at production when there’s a problem to continuous validation and curiosity

Junior engineers can become effective debuggers when proper observability tooling exists - expertise is no longer tied to system familiarity

The industry is moving towards making complex systems tractable rather than just monitoring for problems

Data model choice significantly impacts costs - using the right tool for the right use case is crucial

Platform engineering is bringing product discipline to infrastructure, changing how teams approach observability and system management