Is It Time To Version Observability? (Signs Point To Yes) • Charity Majors • GOTO 2024

Observability 2.0 represents a shift from multiple data sources (metrics, logs, traces) to a single source of truth using structured logs, enabling better debugging and system understanding

High-cardinality data and wide structured logs are crucial for modern observability - collecting rich context about each request provides much more value than traditional metrics-only approaches

The cost model of traditional observability (1.0) becomes exponentially expensive as systems scale, while Observability 2.0 provides better value alignment with engineering needs

Engineering teams should focus on fast feedback loops - getting code into production quickly and safely is key, supported by good observability tools and practices

Canonical logs (one structured log event per request per service) are more powerful than scattered logging throughout code

Modern systems are too complex to debug through mental models and intuition alone - we need tools that preserve context and allow exploration of data

Team performance and system understanding are more important than individual debugging heroics - good observability democratizes the ability to understand systems

Structured data is essential in 2024 - unstructured logs and pure metrics aren’t sufficient for modern debugging needs

Feature flags and separating deploys from releases enables safer, faster software delivery

Engineers should own their code in production and observe how it actually behaves, rather than considering their job done after writing tests