How not to practice observability
summary
Claims that monitoring ≠ observability and that sampling choice matters: five-minute sampling hides intermittent spikes (example: about 200 spikes missed in the scenario) while ten-second sampling drives storage and SaaS cost and overprovisioning. Recommends measuring dashboard usage to avoid technical debt, linking metrics across layers to cut MTTR, tuning alerts to control noise-to-signal ratio, avoiding centralized access bottlenecks, building data-unification platforms with HA and failover, and remembering that tools require process and people changes; ManageEngine and Zoho cite multi-product, multi-customer experience.
AI-written summary · about
chapters (9)
- 0:00Introduction to observability
- 1:02Monitoring vs observability
- 2:55Sampling and data management
- 6:13Strategic dashboard design
- 8:31Holistic incident handling
- 10:17Operational configuration
- 12:21DevOps culture and access
- 14:23System reliability and scale
- 15:55People, tools, and conclusion
related talks
38mThe Ticking Time Bomb of Observability Expectations
39mThe Ticking Timebomb of Observability Expectations
32mRules of Observability: The Data Edition
29mObservability; a gentle introduction
38mThe Murphy's Laws of Observability
40mDemystifying Observability and Monitoring
20mI Can Sell You Observability, But You Can't Buy It
52mObservability: it's not just an ops thing