Re: [Add] Proposed charter and BoF request for IETF 106

[Paul Wouters <[e-mail]>]

On Wed, 9 Oct 2019, Jim Reid wrote:

>> On 9 Oct 2019, at 19:04, tirumal reddy <[e-mail]> wrote:
>>
>> DHCP is not a secure way to discover the local DoT/DoH server.
>
> That may well be true. However that debate belongs on another thread and maybe even another list.
>
> I thought we were meant to be discussing the proposed charter for a new WG and a possible BoF in Singapore. Could we please focus on that?

It is relevant, because if we believe the only thing that is needed is
a DHCP option, then we do not need to spin up a working group. In fact,
I would expect whoever wants to spin up a new group to have done some
preliminary work in this area so that it is clear there is a real large
enough problem to be solved using a WG. I don't get the feeling this
happened. What I see is people looking for a business model for "secure
DNS" services. I don't think that business model needs protocol level
support from the IETF (because as I stated earlier, consumer trust
doesn't work by signed XML statements).

Compare this with starting up a WG to eavesdrop on TLS 1.3 traffic.
Look at how PATIENT hasn't seen any discussion since July 2018. We
didn't start a WG for that either. Why should we do one on eavedropping
encrypted DNS?

As for the "DHCP is not secure" argument. There is absolutely no
expectation of validatable security when someone joins a random
network. When I join a known preconfigured enterprise network, I already
have trust anchors. When I decide to trust a personal DoH server that is
not local, I already have the trust anchors I need as well. An insecure
DHCP option is useful for the enterprise/parental use case, where just
being on the network is a veriably secure action already.

Paul