Re: [Starlink] fiber IXPs in space

[Sebastian Moeller <moeller0@gmx.de>]

Hi David,


> On Apr 18, 2023, at 09:46, David Fernández via Starlink <starlink@lists.bufferbloat.net> wrote:
> 
> PEPs have been mentioned as an example of so called stealth optimization.
> 
> Another example, I think it is IGMP snooping:
> https://en.wikipedia.org/wiki/IGMP_snooping

	I think this is different from "PEPs" (I really dislike the marketing naming... IGMP snooping arguably deals to deal with the fact that IGMP has a different idea of unicast/multicast scoping than is optimal for switched L2 network domains... it is also as far as I can tell opt-in in that one needs to activate it in once's L2.5 devices first. I am not sure whether norrmal geo-stationary internet users can opt-out of the TCP shenanigans played by PEPs?

> So, well, maybe this so called DNS stealth optimization is not so bad,
> if it really is easy to implement and it brings benefits (RTT by
> half), but pros and cons should be carefully evaluated.

	I heartily disagree, stealth DNS "optimization" is not something an ISP should be caught doing behind their users backs. I remember when my former ISPs insisted upon capturing DNS queries to non-existent domains to an advertisement page of their own, I was neither impressed nor happy (even though they did offer an opt-out).
Now, if a hypothetical starlink offer would do that DNS snooping only for DNS queries directed against starlink's own DNS server IP addresses that would be palatable from a who handles the query perspective (starlink in both cases), but from a layering violation perspective this still seems rather vile. If the want to offer DNS forwarders in space, they should simply do so overtly and not high-jack packets directed to a different server.

At least that is my subjective take on this issue.
Regards
	Sebastian

> 
> Regards,
> 
> David
> 
> 2023-04-17 21:00 GMT+02:00, David Lang <david@lang.hm>:
>> On Mon, 17 Apr 2023, Rodney W. Grimes wrote:
>> 
>>>> On Sun, 16 Apr 2023, David Fern?ndez via Starlink wrote:
>>>> 
>>>>> The idea would be that the satellite inspects IP packets and when it
>>>>> detects a DNS query, instead of forwarding the packet to ground
>>>>> station, it just answers back to the sender of the query.
>>>> 
>>>> This would be a bad way to implement it. You don't want to override
>>>> queries to
>>>> other DNS servers, but it would be very easy to create an anycast address
>>>> that
>>>> is served by the satellites.
>>> 
>>> Yes, and the later is what I proposed, the idea of intercepting
>>> someone ELSE'S anycast address and processing it would be
>>> wrong in many ways, in effect a Man In the Middle attack
>>> as stated else where.
>> 
>> I was assuming that it would be done in coordination with the existing user,
>> not
>> as a stealth optimization. I should have made that clear.
>> 
>> David Lang
>> 
> _______________________________________________
> Starlink mailing list
> Starlink@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/starlink