Starlink has bufferbloat. Bad.
 help / color / mirror / Atom feed
From: George Burdell <gb@teklibre.net>
To: Ulrich Speidel <ulrich@cs.auckland.ac.nz>
Cc: starlink@lists.bufferbloat.net
Subject: Re: [Starlink] Some comparison data from New Zealand
Date: Sat, 18 Sep 2021 03:44:06 +0000	[thread overview]
Message-ID: <20210918034406.GA30094@mail.taht.net> (raw)
In-Reply-To: <008dae7a-6c88-1abf-7281-2b6f189fe5c7@cs.auckland.ac.nz>

This paper went by again today, which compared real fiber latencies
around the world against an ideal c-latency.

https://arxiv.org/pdf/1505.03449.pdf

On Fri, Aug 13, 2021 at 10:52:59PM +1200, Ulrich Speidel wrote:
> Just to make sure we're fairly comparing technologies rather than the bad
> shape regulation has left their delivery in North America in... Here are
> some speedtest.net measurements from today (Friday New Zealand time), using
> a selection of servers around the world (with an unashamed selection bias
> towards the Pacific):
> 
> 1) Between 5 and 6 pm from my docked laptop at work (UoA), on GbE with fibre
> backhaul:
> 
> Location 	Server 	Down 	Up 	RTT 	
> Auckland 	WorldNet 	831.45 	914.17 	1 	
> Auckland 	Vodafone 	881.13 	930.93 	3 	
> Sydney 	Optus 	904.4 	335.73 	29 	
> Sydney 	Vodafone Hutchinson 	803.63 	683.94 	27 	
> Hong Kong 	HKIX 	101.04 	73.39 	284 	
> Hong Kong 	i3D.net 	86.42 	39.58 	174 	
> Tokyo 	OPEN Project (via 20G SINET) 	26.31 	30.51 	256 	
> Tokyo 	i3D.net 	326.54 	15.69 	140 	
> Los Angeles 	i3D.net 	342.85 	149.84 	158 	
> Los Angeles 	2degrees 	388.95 	187.5 	125 	
> Rarotonga 	Vodafone Cook Islands 	75.1 	0.31 	280 	
> Apia, Samoa
> 	Digicel Samoa 	736.72 	289.8 	76 	
> Bairiki, Kiribati
> 	ATHKL 	20.41 	3.07 	333 	on satellite (SES/O3b MEO)
> New York 	Atlantic Metro 	481.01 	120.29 	197 	
> London 	UK Dedicated Servers 	114.37 	83.21 	272 	
> Frankfurt 	Deutsche Telekom 	194.7 	77 	299 	
> 
> 2) Between 8:30 and 9:30 pm from my residential home fibre connection (bog
> standard suburban Auckland install). I did a traceroute on a couple of them
> to show that the path isn't always obvious or anywhere near shortest. A lot
> of traffic that should really stay on the western side of the Pacific makes
> its way all the way to the US mainland:
> 
> Location 	Server 	Down 	Up 	RTT 	
> Auckland 	WorldNet 	868.67 	476.85 	3 	
> Auckland 	Vodafone 	894.84 	458.47 	2 	own ISP
> Sydney 	Optus 	713.48 	109.84 	31 	
> Sydney 	Vodafone Hutchinson 	528.21 	166.57 	33 	
> Hong Kong 	HKIX 	119.29 	26.84 	266 	traceroute goes via Hurricane Electric
> in LAX
> Hong Kong 	i3D.net 	120.66 	7.75 	265 	also via HE in LAX
> Tokyo 	OPEN Project (via 20G SINET) 	130.87 	38.63 	235 	
> Tokyo 	i3D.net 	61.02 	31.18 	241 	
> Los Angeles 	i3D.net 	295.05 	65.19 	128 	
> Los Angeles 	2degrees 	544.75 	59.69 	127 	
> Rarotonga 	Vodafone Cook Islands 	43.99 	1.09 	248 	These guys are on the
> Manatua cable from Samoa to Tahiti. Going by the traceroute, the traffic
> goes from NZ to Hurricane Electric in LAX (either via Hawaiki or Southern
> Cross Cable)and from there down on the Hawaiki cable to American Samoa, from
> there to Western Samoa and onto the Manatua cable from there. Must have a
> chat to their CEO to find out what is going on there.
> Apia, Samoa
> 	Digicel Samoa 	365.01 	7.45 	79 	I'm not sure what is going on there with
> their inbound connection. Tide out?
> Bairiki, Kiribati
> 	ATHKL 	12.49 	2.04 	489 	on satellite (SES/O3b MEO)
> New York 	Atlantic Metro 	412.56 	0 	230 	no upload for some odd reason
> London 	UK Dedicated Servers 	45.39 	31.9 	262 	
> Frankfurt 	Deutsche Telekom 	183.99 	23.36 	270 	
> 
> Rates in Mb/s, RTT in ms.
> 
> So working from home's just marginally worse in terms of Internet here in
> Auckland than being at uni. And they haven't even connected the Southern
> Cross Next cable yet. I pay about US$73 / month, which includes a static
> public IPv4 address and home phone.
> 
> As for the islands, the only satellite-connected place with a speedtest
> server I could find was Bairiki, which is part of Tarawa Atoll in Kiribati,
> where one of the ISPs uses SES/O3b MEO (the rest use GEO). It still has a
> whoppingly large RTT (probably around 130-140 ms pure MEO link, without
> queue sojourn times). I've also observed dozens of milliseconds of delay
> INSIDE Tarawa in the past, so quite whether there's a segment of old coconut
> telegraph in there that hasn't been upgraded yet, I don't know. People on
> GEO in this part of the world would be well advised not to host speedtest
> servers - it would eat through their data caps in no time and make them
> deeply unpopular with their fellow locals.
> 
> 
> -- 
> ****************************************************************
> Dr. Ulrich Speidel
> 
> School of Computer Science
> 
> Room 303S.594 (City Campus)
> Ph: (+64-9)-373-7599 ext. 85282
> 
> The University of Auckland
> ulrich@cs.auckland.ac.nz
> http://www.cs.auckland.ac.nz/~ulrich/
> ****************************************************************
> 
> 
> 

> _______________________________________________
> Starlink mailing list
> Starlink@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/starlink


      reply	other threads:[~2021-09-18  3:44 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-13 10:52 Ulrich Speidel
2021-09-18  3:44 ` George Burdell [this message]

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=20210918034406.GA30094@mail.taht.net \
    --to=gb@teklibre.net \
    --cc=starlink@lists.bufferbloat.net \
    --cc=ulrich@cs.auckland.ac.nz \
    /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