Field notes
Practical notes from response-time and uptime work — written for the people who own the systems.
-
Reading p95 response time without panicking the wrong team
A high p95 is not always an application bug. Here is how we separate edge delay, dependency wait, and genuine compute cost.
-
Uptime windows that match how people actually use the app
Calendar-month availability hides overnight maintenance that never touched a real user. Align the clock with usage.
-
Which infrastructure signals earn a permanent place on the wall
More metrics rarely mean clearer diagnosis. Keep the few that explain response time and uptime decisions.
-
How to prepare for a response-time assessment week
Access, journey lists, and freeze windows decide whether the measurement week produces usable baselines.
