Latency budgets are operational tools
A useful budget names where time is spent, how it is observed and who can change it.
Field notes · Network engineering
Make the network
observable.
Practical research on latency, resilience and protocol behavior in real-world networks. Written for people who operate systems, not just diagrams.
↓Recent notes
2026 / 03
Measure recovery, not only uptime
Availability says little about the operator’s experience during the thirty minutes that matter most.
Protocol defaults and the shape of risk
Defaults are hypotheses. Keep them until evidence shows which constraint your network actually has.
Method
Observe.
Change one thing.
Measure again.
Our work favours reproducible tests, reversible changes and boring recovery paths. Performance without a rollback plan is an unfinished experiment.