Starlink has bufferbloat. Bad.
 help / color / mirror / Atom feed
From: Inemesit Affia <inemesitaffia@gmail.com>
To: Dave Taht via Starlink <starlink@lists.bufferbloat.net>
Subject: [Starlink] Re: Ookla's Speedtest results now map the city & PoP where your traffic joins the IXPs as the aircraft moves
Date: Fri, 23 Oct 2026 12:09:11 +0100	[thread overview]
Message-ID: <2e8825ae-a337-44d6-8b59-8ead59104102@gmail.com> (raw)
In-Reply-To: <CAJUtOOgpC_R1DNrBvFoRddY4JObSabeLi08EgQn+ZrsGK8c3Cg@mail.gmail.com>

SpaceX really needs to stop moving US hurricane electric traffic through Oman.

I'm sure AS8529 has BGP no-export communities.

And it's not just there.

Traffic from Seattle to AT&T 7018 goes through Mexico.

Oct 23, 2026 9:13:20 AM Frantisek Borsik via Starlink <starlink@lists.bufferbloat.net>:

> Nice job, Ookla (or rather, Accenture now, lol)
> 
> Kudos to Jianping Pan and Jinwei Zhao (University of Victoria).
> 
> Original post shared by Luke Kehoe annoucing it on LinkedIn:
> https://www.linkedin.com/feed/update/urn:li:activity:7519293982208761856/
> 
> *TL; DR;* Now we can trace Starlink packets from the ground (thanks to work
> done by Jianping Pan and Jinwei Zhao at University of Victoria - see their
> research paper below) AND also from inside the cabin, using passengers’ own
> Speedtest results, thanks to Ookla.
> 
> If you don't want to go there, here is a copy+paste:
> 
> On a Starlink-connected flight, the city and PoP where your traffic joins
> the internet changes as the aircraft moves, and passengers’ own Speedtest
> results now let us map that from inside the cabin across more than 20
> airlines around the world, building on research by Jinwei Zhao and Jianping
> Pan at the University of Victoria (who identified more than 900 Starlink
> aircraft from the ground).
> 
> *Speedtest app results carry a traceroute, and on Starlink those that get a
> reply cross Starlink’s routers at the point of presence (PoP) where traffic
> meets the internet. We matched those routers to PoPs using Starlink’s own
> address records (archived by the UVic team), keeping only passengers with a
> fresh GPS fix at airliner speed (April to September 2026).*
> 
> Over the North Atlantic, passengers connected through Montreal or New York
> until roughly halfway across, then London or Madrid (i.e., directionally
> consistent with the mid-ocean handovers UVic observed).
> 
> Hawaii has Starlink ground stations but no PoP, so passengers within 1,500
> km of Honolulu reached the internet in Los Angeles, nearly 4,000 km away,
> at a median multi-server latency of 95 ms (against 43 ms within 750 km of
> their PoP).
> 
> The most policy-relevant (and interesting) observation here concerns
> in-country egress. Starlink places home users in Oman on its Muscat PoP and
> in Indonesia on Jakarta, Surabaya or Batam, yet passengers over Oman and
> Indonesia connected through Doha and Singapore respectively in our data
> (small numbers, but UVic saw the same and hypothesise these PoPs exist for
> local egress rules). As more governments mandate in-country egress, how
> overflying aircraft are treated is an open question.
> 
> Separately, but relevant, passengers measured about twice the median
> download speed of Starlink homes in the same areas and months in our data.
> 
> The University of Victoria research was able to follow one aircraft for a
> whole flight by combining publicly reachable IPv6 addresses, MPLS routing
> labels and flight tracking data (e.g., one Qatar Airways flight from Doha
> to New York connected through seven cities in turn, at times over 2,000 km
> away, as visualised in the film), and the team found aircraft unreachable
> over India and Pakistan.
> 
> *The research also showed how Starlink's new "anycast users" feature allows
> aircraft to retain the same public IP address as traffic moves between
> geographically distant PoPs, avoiding the TCP connection disruptions
> associated with previous handovers.*
> 
> Research paper linked in the comments.
> 
> The research by Jinwei Zhao and Jianping Pan (University of Victoria),
> *“Measuring
> Starlink Aviation Around the World”*, ACM IMC 2026, open access:
> 
> doi.org/10.1145/3777912.3839781
> 
> *Their method works from the ground.* They identified more than 900
> Starlink aircraft by their router addresses, read Starlink’s own routing
> labels from a Starlink terminal in Canada, and measured latency from 27
> cloud servers over five months. *Our view complements theirs from inside
> the cabin, using passengers’ own Speedtest results.*
> 
> 
> All the best,
> 
> Frank
> 
> Frantisek (Frank) Borsik
> 
> 
> *In loving memory of Dave Täht: *1965-2025
> 
> https://libreqos.io/2025/04/01/in-loving-memory-of-dave/
> 
> 
> https://www.linkedin.com/in/frantisekborsik
> 
> Signal, Telegram, WhatsApp: +421919416714
> 
> iMessage, mobile: +420775230885
> 
> Skype: casioa5302ca
> 
> frantisek.borsik@gmail.com
> _______________________________________________
> Starlink mailing list -- starlink@lists.bufferbloat.net
> To unsubscribe send an email to starlink-leave@lists.bufferbloat.net

  reply	other threads:[~2026-10-23 11:09 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-23  8:23 [Starlink] Ookla's Speedtest results now map the city & PoP where your traffic joins the IXPs as the aircraft moves Frantisek Borsik
2026-10-23 11:09 ` Inemesit Affia [this message]
2026-10-23 17:12 ` [Starlink] " Jianping Pan
2026-10-23 17:55   ` Frantisek Borsik
2026-10-23 17:58     ` Jianping Pan

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://lists.bufferbloat.net/postorius/lists/starlink.lists.bufferbloat.net/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=2e8825ae-a337-44d6-8b59-8ead59104102@gmail.com \
    --to=inemesitaffia@gmail.com \
    --cc=starlink@lists.bufferbloat.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox