- From: Watson Ladd <watson@cloudflare.com>
- Date: Tue, 18 Jul 2017 11:43:51 -0700
- To: Ryan Hamilton <rch@google.com>
- Cc: Martin Thomson <martin.thomson@gmail.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>, Piotr Sikora <piotrsikora@google.com>
- Message-ID: <CAN2QdAGGw3+y4yH3OedVNqbU2bYeXS9hyXP8z9qsY2mMGRVfMQ@mail.gmail.com>
On Tue, Jul 18, 2017 at 11:30 AM, Ryan Hamilton <rch@google.com> wrote: > My apologies for starting this thread on Friday and then going on > vacation! I'm catching up now. > > On Fri, Jul 14, 2017 at 10:32 AM, Martin Thomson <martin.thomson@gmail.com > > wrote: > >> I get that you believe that DNS resolution is valuable here, but I >> don't understand your point about "proxy configuration". If you have >> a proxy configured, you generally don't do the DNS lookup directly, so >> are you saying that you would allow the coalescing for a CONNECT >> tunnel? >> > > Yes, I think so. > > >> Can you verify that I understand your point about Alt-Svc: if the >> origin had been previously contacted and Alt-Svc identified an IP >> address that matches the current IP, then that could be a sufficient >> signal. Usually Alt-Svc is a name though, so that wouldn't help. >> > > Sorry, I wasn't very clear. What I means is that if I connection to > www.example.org which presents an *.example.org certificate, I need some > second piece of information before using it for mail.example.com. In the > alt-svc case, if an earlier connection to mail.example.org advertised > alt-svc for www.example.org, then I would be comfortable coalescing the > connection and would not need to do any resolution of mail.example.org. > This doesn't work for our major usecase, which is using an existing connection to one of our customers to send the content of cdnjs.cloudflare.com. If logging a certificate in CT is enough to enable ORIGIN that is fine. > > Similarly, are you saying that CT could assert something that would >> cause you to skip the DNS lookup for a name? >> > > This was more of a straw-man than a concrete proposal, but yes. Folk in > Chrome security thought that this was plausible, though probably required > more thought. > > Or is this list just to make a more general point about raising the >> bar for gaining the ability to use a name, and not something that is >> specific to this particular example of DNS lookups? >> > > Specific to the issue of skipping DNS lookups when deciding to trust a > certificate. > >
Received on Tuesday, 18 July 2017 18:44:20 UTC