Starlink has bufferbloat. Bad.
 help / color / mirror / Atom feed
From: Alexandre Petrescu <alexandre.petrescu@gmail.com>
To: David Lang <david@lang.hm>
Cc: Sebastian Moeller <moeller0@gmx.de>,
	Steven <bufferbloat-lists@steven.honson.au>,
	starlink@lists.bufferbloat.net
Subject: Re: [Starlink] Info on IP country ranges
Date: Thu, 14 Dec 2023 14:27:38 +0100	[thread overview]
Message-ID: <ade7718b-16c6-47e8-a1a4-205e90afd432@gmail.com> (raw)
In-Reply-To: <4rooss86-p114-8qp1-38r8-rn4302n99or4@ynat.uz>


Le 13/12/2023 à 19:27, David Lang a écrit :
> On Wed, 13 Dec 2023, Alexandre Petrescu via Starlink wrote:
>
>> A tunnel within 3GPP network (GTP) is used, among other reasons, to 
>> support mobility.  The 'mobility', among some interpretations, is to 
>> maintain a constant IP address for a moving end user.
>>
>> Surprisingly, the URL 
>> https://support.starlink.com/?topic=1192f3ef-2a17-31d9-261a-a59d215629f4 
>> explains that that kind of mobility is not supported in starlink, 
>> i.e. the end user might get another IP address if going to some other 
>> area.  It is surprising in that in other starlink.com URLs they offer 
>> starlink service for automobiles, and these typically move a lot.  
>> Maybe the starlink-connected automobiles do change their IP addresses 
>> a lot, but the end users dont care that much.
>>
>> To support mobility within a starlink network - maintain constant all 
>> IP addresses in a car - maybe one would try the DHCPv6 CONFIRM 
>> message to try to maintain the same allocated /56 but it another 
>> area.  Maybe the starlink DHCPv6-PD server will satisfy that CONFIRM, 
>> or maybe not.
>>
>> Or maybe there is a need of some other protocol in starlink, or in 
>> user equipment connected to starlink (Dishy, third party router), to 
>> offer that mobility.  But without adding new latency, of course.
>>
>> (this mobility aspect is on topic of the IP country ranges - 
>> cross-border areas would ideally not break connectivity).
>
> There are two types of keeping your address. One is to have the same 
> address all the time, the other is to keep your address during a session.
Fair enough.
>
> Even if you get a public address, it can change from time to time 
> (just like DHCP addresses could on landline ISPs), 

Well, it depends on ISP.  Indeed some wireline ISPs change the address 
of home customers relatively often - on DHCP lease expiry, MAC-less or 
NAI-less DHCP during reboots, etc.  But other landline ISPs do provide 
always the same IP address to customers.  Some times it is even part of 
the signed contract.


> but that doesn't mean that your address will change during a session 
> (i.e. while you are powered up and connected) even while moving around.

Do you mean that the /56 I'd get from starlink with DHCPv6-PD would stay 
the same if I had ongoing sessions, and moved in and out the hexagon 
cells,or even in-out of a teleport coverage? (driving a car on a longer 
distance).

Remark that /56 can not be equated simply to the behaviour of an IPv4 
address.  That /56 means more than just one address and, additionally, 
it is bound to an IPv6 address to which it is routed (either a GUA or a 
LL, a 'next-hop' address).

The maintenance of that /56 constant can happen with or without 
maintenance of the 'next-hop' address constant.

If starlink keeps constant both the 'next-hop' address and the allocated 
/56 during movements to very wide distances (cell-to-cell, 
teleport-to-teleport) then it could some options.

If starlink keeps constant just the /56 allocated to end user, and 
varies the 'next-hop' address attached to it then it would other options.

There are design options.

>
> If you get a public IP, that IP will change like a DHCP address would 
> on a landline ISP, rarely and mostly when equipment at one end or 
> another was restarted, but not every time you do a new satellite handoff

satellite handoff: do you mean the handoff provoked by the movement of 
satellites to a ground-fixed user, or do you mean the movement of end 
user in and out of hexagon cells covered by satellites?

IMHO, I think it can mean either of the two.  For the latter (movement 
of end user between cells) there'd probably be another /56 delegated to 
an end user, regardless of having ongoing sessions - DHCP does not check 
the status of apps.

Further, a very wide area movement might provoke a change in teleport, 
not just a change in hexagon cells.  A change in teleport certainly 
provokes a change in the allocated /56 of the end user, again regardless 
of her having running sessions.  In order to keep a same /56 allocated 
to a same end user handed over from one teleport to another there would 
be a need of some teleport-to-teleport protocol, maybe a tunnelled BGP, 
or other.

Alex


>
> David Lang

  reply	other threads:[~2023-12-14 13:27 UTC|newest]

Thread overview: 48+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-12-04  0:15 Noel Butler
2023-12-04  0:44 ` J Pan
2023-12-04 12:04   ` Noel Butler
2023-12-04 18:17     ` J Pan
2023-12-07 11:54       ` Alexandre Petrescu
2023-12-07 18:07         ` J Pan
2023-12-07 20:10           ` Alexandre Petrescu
2023-12-07 20:40           ` David Lang
2023-12-08  5:57         ` Freddie Cash
2023-12-08  8:30           ` Alexandre Petrescu
2023-12-08  8:37             ` Alexandre Petrescu
2023-12-08 15:27               ` Freddie Cash
2023-12-08 17:26                 ` Alexandre Petrescu
2023-12-08  8:40             ` Steven
2023-12-08 10:22               ` Alexandre Petrescu
2023-12-08 10:31                 ` Steven
2023-12-08 11:48                   ` Alexandre Petrescu
2023-12-08 11:54                     ` Steven
2023-12-08 12:07                       ` Alexandre Petrescu
2023-12-08 12:24                         ` Steven
2023-12-08 12:46                           ` Alexandre Petrescu
2023-12-08 14:15                             ` Alexandre Petrescu
2023-12-11 15:30                           ` Alexandre Petrescu
2023-12-12  1:09                             ` Steven Honson
2023-12-12  2:29                               ` J Pan
2023-12-12  2:43                                 ` Steven
2023-12-12 10:22                                   ` Alexandre Petrescu
2023-12-12 10:33                                     ` Steven
2023-12-12 10:50                                       ` Sebastian Moeller
2023-12-13 10:33                                         ` Alexandre Petrescu
2023-12-13 14:37                                           ` Marc Blanchet
2023-12-13 18:27                                           ` David Lang
2023-12-14 13:27                                             ` Alexandre Petrescu [this message]
2023-12-23 21:35                                               ` J Pan
2024-01-06 15:01                                                 ` Alexandre Petrescu
2024-01-08 10:01                                                   ` [Starlink] starlink topology with tunnels (was: Info on IP country ranges) Alexandre Petrescu
2024-01-08 10:06                                                   ` [Starlink] starlink topology with tunnels Alexandre PETRESCU
2023-12-12 10:52                                       ` [Starlink] Info on IP country ranges Alexandre Petrescu
2023-12-08 10:28               ` Alexandre Petrescu
2023-12-08 10:31                 ` Gert Doering
2023-12-13 20:58 David Fernández
2023-12-13 22:57 ` Marc Blanchet
2023-12-13 22:59   ` Marc Blanchet
2023-12-24  1:29   ` Lukasz Bromirski
2024-01-06 14:59     ` Alexandre Petrescu
2024-01-06 15:06       ` Dave Taht
2024-01-08 10:54         ` Sebastian Moeller
2024-01-08 13:16           ` Alexandre Petrescu

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=ade7718b-16c6-47e8-a1a4-205e90afd432@gmail.com \
    --to=alexandre.petrescu@gmail.com \
    --cc=bufferbloat-lists@steven.honson.au \
    --cc=david@lang.hm \
    --cc=moeller0@gmx.de \
    --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