Hello from the other side
summary
Deployments wake people at night or blow up datastores; measurable fixes matter. The talk shows DNS DNS TTL affecting rollout speed, compares HTTP/1.1 vs HTTP/2 performance, points to MySQL connection limits and ProxySQL-style multiplexing, and recommends tracing, structured events and Honeycomb for observability. A practical canonical logging field set (host, cluster, user, release, request_id, timestamp) plus minimal metrics — request rate error rate duration — turns noise into actionable alerts and makes rollbacks, ownership and incident triage tractable.
AI-written summary · about
chapters (7)
- 0:00Start
- 2:13SOFTWARE ENGINEER
- 11:27WHAT'S THE ADDRESS?
- 29:38TRACING LOGS / EVENTS METRICS
- 38:36METRICS (FATIGUE)
- 40:54METRICS (WHAT?)
- 43:10CREATE CHANGE
