Re: Structuring the BKK spin bit discussion

[Ted Hardie <[e-mail]>]

Just to add one additional piece of data here that was the basis of a lot
of discussion in the design team:  consider whether the handshake alone
discloses the data you are trying to protect.  For the case where you are
trying to obscure a hidden path segment like a VPN, a comparison of
handshake time to expected time is almost certainly enough.

regards,

Ted

On Mon, Oct 29, 2018 at 3:58 PM Brian Trammell (IETF) <[e-mail]>
wrote:

> hi Martin, Christian, all,
>
> > On 29 Oct 2018, at 23:29, Martin Thomson <[e-mail]>
> wrote:
> >
> > On Tue, Oct 30, 2018 at 3:54 AM Christian Huitema <[e-mail]>
> wrote:
> >> I think the strongest objection to the spin bit was put up by Marten
> during the last interim: measuring the RTT with the spin bit discloses the
> use of hidden path segments like VPN. This issue was not discussed during
> the privacy analysis.
> >
> > I had assumed that was part of the analysis and it was covered by the
> > assumption that spinning could be disabled
>
> +1. Probabilistically disabling spinning, which seems necessary if we want
> some grease to help us reserve the right to change the semantics of the bit
> at the spin bit's location in the wire image, should ensure that endpoints
> that want to disable spinning for their own reasons will have a large
> anonymity set to hide in, even in a future with perfect implementation and
> deployment of the spin bit.
>
> >> One solution would be to remove the spin bit from the spec, trading off
> better privacy for worse management. I am considering another solution in
> which privacy sensitive clients hide the RTT by controlling the spin, for
> example spinning at fixed intervals. I plan testing that option in Picoquic.
> >
> > I've done a little thinking about this one, and it might conflict with
> > the natural signals the transport emits, along the lines of what
> > Andrew McGregor has mentioned on a couple of occasions.  If the spin
> > bit is enabled, then privacy-sensitive endpoints will need to make a
> > hard call regarding standing out.
> >
> > Note also that you probably can't hide the fact that you aren't at the
> > same network location as the address you are using.  Spoofing a
> > shorter RTT is impossible in general because you have to assume edges
> > that haven't arrived yet.  If there are no edges you expose the
> > charade.
>
> You cannot reliably hide the fact that you're not at the same network
> location as the address you're using, spin bit or no. Most of the methods I
> know of for detecting people trying to hide where they're coming from don't
> rely on RTT at all. (Yes, I'm a Netflix customer who relies on Hurricane
> Electric's tunnel broker for IPv6 connectivity, why do you ask? ;) )
>
> The recent academic work I'm aware of in this space (Weinberg et al "How
> to Catch When Proxies Lie", to appear at IMC this week) uses minimum RTT
> (with injected active measurement traffic) to conservatively draw exclusion
> circles to show that VPN providers that promise exits in certain countries
> probably don't actually deliver them. Not being vulnerable to this kind of
> exclusion analysis requires that you actually inject latency, not just in
> the spin bit, but over the entire flow (including the handshake).
>
> RTT *higher* (even much higher) than an expected range for a given address
> pair isn't reliable enough to derive any sort of verdict from, due to the
> nature of delay and delay variability in the Internet.
>
> So while I guess you can lie with the spin bit (even using rudimentary
> signal processing magic to take an assumed real RTT and make it look like a
> given lower target RTT, as long as you're not too worried about getting
> caught when your assumptions fail to hold), it's probably neither necessary
> nor sufficient to do so to hide a VPN exit. It's cheaper and more effective
> to just turn it off and hope.
>
> Cheers,
>
> Brian
>
>