Leveraging Data Output for Architectural Optimization
summary
Argues that effective cloud-native architecture comes from picking the right telemetry, not hoarding it: use metrics before logs for fast detection, combine traces with logs to localize and explain failures, and tier storage to avoid runaway costs. Recommends standardizing formats with OpenTelemetry standardizes telemetry and forming an observability CoE to enforce it. Notes modern scale — from ~17 servers to 1,700 containers — and says AI/ML and generative tools accelerate correlation and troubleshooting.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:27Distributed systems complexity
- 3:56Managing distributed data
- 9:36Data types metrics traces logs
- 11:45Common data handling mistakes
- 16:44Hot storage costs retention
- 20:37OpenTelemetry standardization
- 23:48Observability practice
- 25:53AI for telemetry analysis
- 29:22Summary
related talks
32mThe Role of Observability in Cloud Native Environments
31mCharting New Frontiers: The Evolution of Observability in Cloud Environments
35mBeyond Logs, Metrics and Traces. The Data Analytics Paradigm Shift in Observability — Dotan Horovits
15mImportance of Observability in Platform Engineering
13mNext-Gen Observability: reduce MTTR
16mBuilding Observability into Cloud-Native Applications
30mPanel: Telemetry That Matters
12mHarnessing Chaos Engineering to Build Robust Systems