i keep writing blog posts 😔 https://jyn.dev/brownouts-reveal-system-boundaries/
Nice quote starting that post!
The conversations I've had along these lines recently is convincing teams to write up their SLI/SLOs and the idea that improving upon the SLOs of all your dependencies is actually really hard. Out the gate, you're at least as bad as your worst dependency SLO.
Example, we also had an outage due to some interaction with redis DB cache key sizes and feature-flags cardinality. I didn't know the app needed redis to run in production until it happened.
yeah! that's exactly the kind of thing i'm thinking about, you don't know what you depend on until it breaks