Mon 23 Oct 2023

HotNets, HotNets, HotNets

  • Don’t try and rewrite your whole paper a day before the deadline.
  • Got it in, and need to think about travel and accomadation next.
  • The program has been published, and I’m on the first day at 1:15pm - 2:45pm in ‘Session 3: The architecture of the Internet, and datacenter networks’.

SIGCOMM SNS

Magnus and I threw some ideas around on Thurday morning. Apple’s ARKit might be something to look at in lieu of a Vision Pro. We also discussed getting some experience with iPhone’s lidar precision, how to bootstrap a connection, the issues with civic name encodings, and the necessity to get some estimation of load.

Regarding an implementation, once we have a bit more of an idea of what it might look like, github.com/RyanGibb/aeon would be a good starting point.

OCaml Wake-on-Lan

Following up on hibernia, I created an OCaml implementation of the Magic Packet format for the Wake-on-Lan (WoL) standard.

Malte Visit

Spoke to Malte about the Spatial Name System paper. He made some interesting points around the benefit of expressing a concise problem statement. I expressed a multi-problem on:

  1. Local-first connectivity
  2. Low latency resolution
  3. Resolving locations
  4. Interoperate with existing network protocols

Malte thought No. 1 was must compelling using Facebook’s DNS/BGP meltdown resulting in lack of access to physical infrastructure as a motivating case. He was fine with the idea of a global onthology of civic heirarchy of names.

It’s interesting to see different people’s objections to this. Keshav’s was the lack of a succinct problem statement. Ian Lewis’ was the difficulty in defining a global ontology of devices. And Malte’s was in the neccessity of low latency operation.

SRG Seminars

I’m still the de-facto organiser of the SRG seminars as no one was found to replace me. I should send out an email to try and recruit a replacement.

Teaching

  • TA’d L50.