[Bloat] retransmit cost over cellular

Juliusz Chroboczek jch at irif.fr
Mon Sep 18 08:21:59 EDT 2023


Hi Dave!

> https://nickvsnetworking.com/mobile-ipv6-tax/

« This means my Android phone consumes 4.5 MB of cellular data in an hour
  while sitting on the desk, with 16,889 packets in/out. »

So even discounting the headers, the phone receives 70 Commodore C64 worth
of data when idle.  Every freaking hour.

« We have 16,889 packets, 6,417,732 bytes in total, minus 97 bytes from
  each gives us 1,638,233 of headers to drop (~1.6MB) giving us a total of
  4.556 MB traffic to/from the phone itself. »

The average packet size is 269 bytes.  Even if we assume that every second
packet is a pure ACK, that's still on the order of just 500 bytes for data
packets.

Conclusions:

1. The amount of data being received is outrageous, which indicates the
   use of JSON or XML to encode the data.  (See RFC 3252.)  (Just kidding,
   please see RFC 8949 instead.)

2. The packet size is small, which indicates the use of a chatty REST-like
   API rather than a streaming protocol.  The use of streaming has been
   known since at least the 1970s, and well-documented since the 1990s.
   For example, both IMAPv4 and Caldav can do streaming synchronisation
   just fine.

3. The « IPv6 tax » could be reduced by 70% if the packets were reasonably
   sized.  By 90% if the application-layer protocol were efficient enough
   to allow delack to trigger.

Conclusion of the conclusions:

4. The « IPv6 tax » is negligible when compared to the JSON/XML/REST tax.

-- Juliusz


More information about the Bloat mailing list