- From: Kent Watsen <kwatsen@juniper.net>
- Date: Wed, 12 Aug 2015 16:56:52 +0000
- To: Stephen Farrell <stephen.farrell@cs.tcd.ie>, "ietf-http-wg@w3.org Group" <ietf-http-wg@w3.org>
- CC: "Max Pritikin (pritikin)" <pritikin@cisco.com>
[CC-ing Max Pritikin] Hi Stephen, Comments inlined below. Thanks, Kent On 8/12/15, 11:59 AM, "Stephen Farrell" <stephen.farrell@cs.tcd.ie> wrote: > >Hi Kent, > >On 12/08/15 16:33, Kent Watsen wrote: >> >> Trusted >> Internet-based Deployment-specific >> Device Bootstrap Server Bootstrap Server >> | | | >> | | | >> | HTTPS using factory | | >> | default trust anchor | | >> |------------------------->| | >> | | | >> | HTTP redirect, with | | >> | deployment-specific | | >> | trust anchor | | >> |<-------------------------| | >> | | | >> | | >> |HTTPS using learned trust anchor | >> |----------------------------------------------->| >> | | >> >> >> >> FWIW, this is considered secure, as the trust anchor is learned through >>a >> trusted connection. > >Eh... considered secure by whom? I would not consider >that "secure" if any widget-vendor from whom I purchase >anything can then pretend to be any web site. Well, both Max and I felt it was OK for the bootstrapping use case. Admittedly, it has not been thoroughly reviewed yet...though this email thread is taking it on now :) The widget-vendor isn't pretending to be another website - it is redirecting the client to another website. For instance: Request (assuming device's serial number is "123456") ------- GET https://widget-vendor.com/bootstrap-server:devices/device=123456 Response -------- HTTP/1.1 200 OK Content-Type: application/xrd+xml Content-Length: nnn Location: https://rightful-owner.com/bootstrap-server:devices/device=123456 <X.509v3 CA certificate for rightful-owner.com here> >Do you really mean trust-anchor here? Or perhaps you >mean "some kind of special public key only to be used >for the specific URL that is in this 30x response and >only for 1 (really 1?) time"? My terminology may be incorrect, but as the above example implies, the X.509v3 cert returned is expected to be subsequently used to verify the HTTPS server (rightful-owner.com). This in lieu of relying a global CA certificates, such as are distributed by web browsers. >And what if that URL is say bigbank.example.com? How >is that to not work? I think your question regards the general applicability of this idea by web browsers, where having the web browser dynamically learn a new trust anchor certificate, even if over a trusted connection, can lead to misuse. Is that right? - that is, is your concern is for generic use more so than the specific use of zerotouch bootstrapping? Thanks again, Kent
Received on Wednesday, 12 August 2015 16:57:21 UTC