
Three Pillars, Internal State, and Monitoring

#88 - Observability Engineering - Liz Fong-Jones
Topic 3 of 12
Published May 16, 2022 · Clip 10:08–15:07 (4:59)
Liz pushes back on the 'three pillars' narrative, arguing that one or two high-quality signals suffice, metrics are pre-aggregated and limiting, and new signals like continuous profiling exist. She distinguishes observing internal state (what the code was doing) from external state (CPU/memory utilization). She explains that observability is a superset of monitoring: monitoring tells you something broke but not why, while Service Level Objectives are 'monitoring 2.0' that differentiate expected from unexpected error levels using error budgets, avoiding 2 AM pages for flapping alerts.