Connecting the dots: observability in large e-commerce environments
summary
A real-world observability path: a real-time CDN logs pipeline (Akamai → processing) groups requests into pages and calculates error rates so teams see which pages fail; rolling out 100% distributed tracing across production takes about six months, exposes hidden long-tail failures (one traced to an obscure load-balancer queue setting causing 5s timeouts), and surfaces correlation between backend CPU, DB wait and user-visible errors; open-source options (Jaeger/Zipkin) work, proprietary APMs are often much more expensive.
AI-written summary · about
chapters (8)
- 0:00Start
- 7:40What Is the Difference between Monitoring and Observability
- 8:10Observability
- 11:02Three Pillars of Logging
- 17:15Distributed Tracing
- 29:53How Long Would You Say It Takes at a Minimum To Get Started with Distributed Tracing
- 34:16What Would Be Your Recommendation for a Sampling Rate When Using Distributed Tracing
- 35:38Performance Impact
related talks
01h 07m"Adding Observability to Distributed Systems"
01h 02mAdding Observability to Distributed Systems
38mTwo Roads Diverged in the Observability Woods: I Took the E2E One
38mOff the Beaten Path: Moving Observability Focus from Your Service
16mBuilding Observability into Cloud-Native Applications
49mImproving Observability of Distributed Systems
56mFull-stack observability and application monitoring with
01h 01mBetter reliability and uptime with analytics and observability