[Cake] starlink testing

[Cake] starlink testing

[Dave Taht]

I find myself in front of a starlink terminal this morning, and have
been doing various tests. The behavior of their network is like
nothing I've ever seen before, appearing to re-adjust the available
bandwidth on a 10sec interval, and... yes... they have bufferbloat...
bad.

Still running quite the large battery of tests at the moment. They are
behind a large cgnat, and
no ipv6 is available, either. I can make available some tests and
packet caps if anyone wants a look.

If anyone would like to drop in on me and help talk me
down^H^H^H^H^H^H^H strategize on how
best to tell 'em about things like fq_codel and cake, I can be found
over here (any login you want, no password) from 7AM PDT through 9AM.

https://tun.taht.net:8443/group/bufferbloat



--
"For a successful technology, reality must take precedence over public
relations, for Mother Nature cannot be fooled" - Richard Feynman

dave@taht.net <Dave Täht> CTO, TekLibre, LLC Tel: 1-831-435-0729

Re: [Cake] [Bloat] starlink testing

[David Lang]

[-- Attachment #1: Type: text/plain, Size: 1638 bytes --]

are you able to sniff between the router and the dish? I'm curious how much of 
the smarts is in the dish vs the router. My hope is that the router is just a 
conventional router with the satellite network smarts in the dish.

David Lang


On Mon, 19 Apr 2021, Dave Taht wrote:

> Date: Mon, 19 Apr 2021 06:56:33 -0700
> From: Dave Taht <dave.taht@gmail.com>
> To: bloat <bloat@lists.bufferbloat.net>,
>     Cake List <cake@lists.bufferbloat.net>
> Subject: [Bloat] starlink testing
> 
> I find myself in front of a starlink terminal this morning, and have
> been doing various tests. The behavior of their network is like
> nothing I've ever seen before, appearing to re-adjust the available
> bandwidth on a 10sec interval, and... yes... they have bufferbloat...
> bad.
>
> Still running quite the large battery of tests at the moment. They are
> behind a large cgnat, and
> no ipv6 is available, either. I can make available some tests and
> packet caps if anyone wants a look.
>
> If anyone would like to drop in on me and help talk me
> down^H^H^H^H^H^H^H strategize on how
> best to tell 'em about things like fq_codel and cake, I can be found
> over here (any login you want, no password) from 7AM PDT through 9AM.
>
> https://tun.taht.net:8443/group/bufferbloat
>
>
>
> --
> "For a successful technology, reality must take precedence over public
> relations, for Mother Nature cannot be fooled" - Richard Feynman
>
> dave@taht.net <Dave Täht> CTO, TekLibre, LLC Tel: 1-831-435-0729
> _______________________________________________
> Bloat mailing list
> Bloat@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat

Re: [Cake] [Bloat] starlink testing

[Michael Richardson]

[-- Attachment #1: Type: text/plain, Size: 408 bytes --]


David Lang <david@lang.hm> wrote:
    > are you able to sniff between the router and the dish? I'm curious how much
    > of the smarts is in the dish vs the router. My hope is that the router is
    > just a conventional router with the satellite network smarts in the dish.

No.
See the teardowns, such as: https://www.youtube.com/watch?v=QudtSo5tpLk
It's a huge synthetic antenna.  It's openwrt though.


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: [Cake] [Bloat] starlink testing

[David Lang]

I've watched those, and it's PoE communication to the dish, and there is a 
processor on the dish with a serial console.

I haven't seen any follow-ups where anyone has gotten a login to the dish or has 
sniffed the traffic between the dish and the router.

I haven't seen anything that would show that the router has (or doesn't have) 
any awareness of the satellite network.

That's why I was asking about sniffing between the devices to see what's 
happening there.

David Lang

On Mon, 19 Apr 2021, Michael Richardson wrote:

> David Lang <david@lang.hm> wrote:
>    > are you able to sniff between the router and the dish? I'm curious how much
>    > of the smarts is in the dish vs the router. My hope is that the router is
>    > just a conventional router with the satellite network smarts in the dish.
>
> No.
> See the teardowns, such as: https://www.youtube.com/watch?v=QudtSo5tpLk
> It's a huge synthetic antenna.  It's openwrt though.
>
>

Re: [Cake] [Bloat] starlink testing

[Michael Richardson]

David Lang <david@lang.hm> wrote:
    > I haven't seen any follow-ups where anyone has gotten a login to the dish or
    > has sniffed the traffic between the dish and the router.

True, I haven't heard, but I haven't been looking.

    > I haven't seen anything that would show that the router has (or doesn't have)
    > any awareness of the satellite network.

With multi-hundred antenna to be steered, and an AMD64 CPU on board, I can't
imagine that that CPU isn't doing some significant amount of DSP.
If they had offboarded all the analog parts to another DSP or discrete logic,
then probably they wouldn't need a 64-bit CPU onboard: they could have used
something slower/cheaper.  Well, maybe there is future proofing involved.

I'm very disappointed that "Something simpler than IPv6" turned out to be "CGN".

--
]               Never tell me the odds!                 | ipv6 mesh networks [
]   Michael Richardson, Sandelman Software Works        |    IoT architect   [
]     mcr@sandelman.ca  http://www.sandelman.ca/        |   ruby on rails    [

Re: [Cake] [Bloat] starlink testing

[Dave Taht]

I'm still busy doing more and more sophisticated tests and packet
captures. I am inclined to "go dark" for a while whilst I do that, but
if anyone would like to help re-write the ultimate blog entry, or
help out on instrumenting more tests, please contact me offlist.

Apparently there is IPv6 under test... somewhere... according to
reddit. Tunnelling my ipv6 via wireguard "just worked", so there's
that. There have been a few other surprises, one rather major.

As much as I dislike the cgnat I don't see how starlink had any other
choice, and the layer below that is hopefully capable of carrying
ipv6 well.

On Tue, Apr 20, 2021 at 7:10 AM Michael Richardson <mcr@sandelman.ca> wrote:
>
> David Lang <david@lang.hm> wrote:
>     > I haven't seen any follow-ups where anyone has gotten a login to the dish or
>     > has sniffed the traffic between the dish and the router.
>
> True, I haven't heard, but I haven't been looking.
>
>     > I haven't seen anything that would show that the router has (or doesn't have)
>     > any awareness of the satellite network.
>
> With multi-hundred antenna to be steered, and an AMD64 CPU on board, I can't
> imagine that that CPU isn't doing some significant amount of DSP.
> If they had offboarded all the analog parts to another DSP or discrete logic,
> then probably they wouldn't need a 64-bit CPU onboard: they could have used
> something slower/cheaper.  Well, maybe there is future proofing involved.
>
> I'm very disappointed that "Something simpler than IPv6" turned out to be "CGN".
>
> --
> ]               Never tell me the odds!                 | ipv6 mesh networks [
> ]   Michael Richardson, Sandelman Software Works        |    IoT architect   [
> ]     mcr@sandelman.ca  http://www.sandelman.ca/        |   ruby on rails    [
>


-- 
"For a successful technology, reality must take precedence over public
relations, for Mother Nature cannot be fooled" - Richard Feynman

dave@taht.net <Dave Täht> CTO, TekLibre, LLC Tel: 1-831-435-0729

Re: [Cake] [Bloat] starlink testing

[Michael Richardson]

Dave Taht <dave.taht@gmail.com> wrote:
    > As much as I dislike the cgnat I don't see how starlink had any other
    > choice, and the layer below that is hopefully capable of carrying
    > ipv6 well.

IPv6 first, with NAT64 would have been significantly better and easier.