Measuring packet loss without guessing
Most "the network is slow" reports are really MTU problems. A short walk through path MTU discovery, ICMP black holes and why clamping MSS at the edge saves an afternoon.
Short, practical write-ups about backend systems, networking and the boring infrastructure that keeps things running. Updated when something is worth writing down.
Most "the network is slow" reports are really MTU problems. A short walk through path MTU discovery, ICMP black holes and why clamping MSS at the edge saves an afternoon.
How we keep a fleet of single-purpose binaries deployable: one artifact, one config file, one health endpoint, and no orchestrator in the way.
Structured logs are only useful if you can read them at 3am. Notes on retention, field naming and keeping cardinality under control.
Benchmarks for TLS termination in front of application servers, and when the extra hop is actually free.