Starlink has bufferbloat. Bad.
 help / color / mirror / Atom feed
From: Dino Farinacci <farinacci@gmail.com>
To: Darrell Budic <budic@onholyground.com>
Cc: starlink@lists.bufferbloat.net, lisp-starlink@lispers.net
Subject: Re: [Starlink] starlink and VPN
Date: Wed, 22 Jan 2025 15:42:07 -0800	[thread overview]
Message-ID: <DD9AA5FB-4ECD-430C-8623-C849773F2E19@gmail.com> (raw)
In-Reply-To: <9406D818-99D0-40BC-BC58-4AC4F778D7EC@onholyground.com>

[-- Attachment #1: Type: text/plain, Size: 736 bytes --]

What I was trying to test was the anticipation of ISLs so we could try house to house connectivity and measure latency. I had to implement a "decentralized NAT traversal" function inside the LISP module. We did ping, but at the time I think the packets went "up-down-up-down" rather than "up-over-down".

We also ran a IP multicast chat program over it so we were able to head-end-replicate from a starlink source to 2 starlink group members. See attached slide for the solution. 

Full sides at 

  https://www.dropbox.com/scl/fi/9iwsp7rlh3g9mrhbqqul2/ipv6-gaap-over-lisp-ietf-prague.pdf?rlkey=tzz1vwutq8d7vwm1tqzvnbpfv

GAAP was a dynmic decentralized group address alloation protocol that ran in the gaapchat app.

Dino


[-- Attachment #2: PastedGraphic-1.png --]
[-- Type: image/png, Size: 4247921 bytes --]

[-- Attachment #3: Type: text/plain, Size: 3108 bytes --]





> On Jan 22, 2025, at 3:29 PM, Darrell Budic via Starlink <starlink@lists.bufferbloat.net> wrote:
> 
> I’ve run openvpn over starlink on v4 and v6, client on starlink, server in my DC. I have since switched to wireguard and haven’t had any troubles other than having to play some server side tricks to get wireguard to let me bring up more than one tunnel via the starlink CGN. I used some fixed ports and aliased a couple IP addresses and used some snat to force it to see things as different endpoints, otherwise wireguard can’t distinguish the end points on the server. I don’t think openvpn would have that problem.
> 
>  -Darrell
> 
>> On Jan 22, 2025, at 5:25 PM, Dino Farinacci via Starlink <starlink@lists.bufferbloat.net> wrote:
>> 
>> We did test this. IPv6 over LISP. Which means IPv6 EIDs used at transport layer over IPv4 encapsulation. But we did not run  over IPv6 locators.
>> 
>> So what I mean is we tested:
>> 
>> (1) IPv4 overlay on an IPv4 satellite underlay 
>> (2) IPv6 overlay on an IPv4 satellite underlay 
>> 
>> Dino
>> 
>>> On Jan 22, 2025, at 2:53 PM, Dave Taht <dave.taht@gmail.com> wrote:
>>> 
>>> How about openvpn over ipv6?
>>> 
>>> Or with a static ip assigned to the starlink?
>>> 
>>> Wireguard works for me....
>>> 
>>> On Tue, Jan 21, 2025 at 3:02 PM Dino Farinacci via Starlink
>>> <starlink@lists.bufferbloat.net> wrote:
>>>> 
>>>> I haven't tried openvpn but I have done a bunch of testing of LISP over Starlink. If anyone wants details I can point you to an Internet Draft and slides that have been presented a couple of times at IETF.
>>>> 
>>>> Dino
>>>> 
>>>>> On Jan 20, 2025, at 11:25 PM, David Lang via Starlink <starlink@lists.bufferbloat.net> wrote:
>>>>> 
>>>>> has anyone done any work with openvpn over starlink (especially if they got the connectors to completely bypass the router)?
>>>>> 
>>>>> I've got the basic connectivity working, but am having problems trying to get openvpn to work (especially for traffic back through the cgnat to the router on the starlink side)
>>>>> 
>>>>> the logs on the client are reporting link local: (not bound) when trying UDP, when I try TCP (and clamp the mtu low) I can connect from the starlink side (st least sometimes) but cannot get the routing the other way to work
>>>>> 
>>>>> David Lang
>>>>> _______________________________________________
>>>>> Starlink mailing list
>>>>> Starlink@lists.bufferbloat.net
>>>>> https://lists.bufferbloat.net/listinfo/starlink
>>>> 
>>>> _______________________________________________
>>>> Starlink mailing list
>>>> Starlink@lists.bufferbloat.net
>>>> https://lists.bufferbloat.net/listinfo/starlink
>>> 
>>> 
>>> 
>>> -- 
>>> Dave Täht CSO, LibreQos
>> 
>> _______________________________________________
>> Starlink mailing list
>> Starlink@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/starlink
> 
> _______________________________________________
> Starlink mailing list
> Starlink@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/starlink


  reply	other threads:[~2025-01-22 23:42 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-01-21  7:25 David Lang
2025-01-21 11:43 ` David Collier-Brown
2025-01-21 15:12 ` b. angel
2025-01-21 15:22   ` David Lang
2025-01-21 15:27     ` Sebastian Moeller
2025-01-21 15:36       ` Nils Andreas Svee
2025-01-21 15:45       ` David Lang
2025-01-21 16:14     ` Keith Simonsen
2025-01-21 16:45       ` David Lang
2025-01-21 15:36 ` Gert Doering
2025-01-21 15:52   ` David Lang
2025-01-21 23:02 ` Dino Farinacci
2025-01-22 22:53   ` Dave Taht
2025-01-22 23:25     ` Dino Farinacci
2025-01-22 23:29       ` Darrell Budic
2025-01-22 23:42         ` Dino Farinacci [this message]
2025-01-23  3:10     ` David Lang
2025-01-23 17:31       ` J 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=DD9AA5FB-4ECD-430C-8623-C849773F2E19@gmail.com \
    --to=farinacci@gmail.com \
    --cc=budic@onholyground.com \
    --cc=lisp-starlink@lispers.net \
    --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