do bigger packets actually help
FW-001inconclusive
mtu size vs. real-world throughput. the lab answer and the internet answer disagreed.
hypothesis
larger mtu, fewer headers, faster transfers. obvious, measurable, done by lunch.
apparatus
- two machines
- iperf3
- one home router with opinions
procedure
- 1.baseline at mtu 1500. three runs, median kept.
- 2.raise mtu in steps. rerun. note where things fragment.
- 3.repeat across the router, because the internet is not a lab bench.
measurements
lan, mtu 1500SAMPLE VALUE
lan, mtu 9000SAMPLE VALUE
wan, any mtusee failure notes
unexpected result
failure notes
- SAMPLE: what actually broke the clean result (fragmentation? a middlebox? the router?)
conclusion
unwritten. the test continues.
next experiment
repeat with path mtu discovery logging. write down everything.