[Add] DRDP: response to feed backs provided during the meeting

[Daniel Migault <[e-mail]>]

Hi,

Thank you for the feed backs provided during the meeting. Here
is a response to three concerns raised during the meeting. Feel
free to let me know your thoughts.

1. One comment was that having a unique centralize repository of
resolving domains for public resolvers was a terrible idea. I
agree with this as it implies a complex governance. I will update
the draft and let it as "a" repositories for resolving domains
instead. I however believe that repositories of resolving domain
share a common format so one is able to switch easily and move
from one repo to the other.

Note that resolving domains associated to local resolvers may also
happen to be found in such repos. Currently I am not convinced
that we should envision a large number of local resolving
domains, but I am happy to hear otherwise.

2. Another comment concerned the number of resolving domains. I
think I recall that the current list of resolvers used by the
Google Public DNS is greater than 11.000. I do not know if that
list is available but I am wondering whether these entries are IP
addresses or domain names. I suspect that IP addresses are used
and that some ISPs are proposing multiple IP addresses. If that
is correct, handling resolving domains instead of IP addresses
might help reduce the length. Of course this assume sthe
discovery protocol will help discovery the IP addresses. In
addition, I am also wondering how much these entries are "public"
resolvers that are globally available. I suspect that, thought
global IP addresses may be used, their access might be limited to
DNS client that belong to a specific network. Again this is only
a suspicion and I am happy to learn more about.

Thought I am not sure I am using relevant links, the link below
mention a 50ish of domains.
https://github.com/curl/curl/wiki/DNS-over-HTTPS#publicly-available-servers
https://dnsprivacy.org/wiki/display/DP/DNS+Privacy+Test+Servers#DNSPrivacyTestServers-Publicresolvers

The link below provides much more public resolvers 11.000 public
resolving, but I suspect that most of them are open resolver. Of
course these are public, but maybe not what we expect when we
mention public resolver - which is probably why a centralized
repository is a terrible idea.  https://public-dns.info/

Overall, I have the impression that we have 11000 entries because
we are missing a discovery protocol and that a the list for
global resolver (expressed as domain) is expected to remain much
smaller.

3. Regarding the format, to be used to host a list of resolving
domain, the use of DNS messages to host the list seems to me wise
as DNS is the protocol that for sure DNS client will understand.
In addition, I believe that tcp will be able to handle large list
of domains. One drawback thought seems that we may not be able to
compress this list with gzip for example. Using a URI that
redirects to an HTTPS or FTP requires the DNS client to have
these protocols which is not given nor necessary. I am happy to
hear your thoughts on this.

Yours,
Daniel



-- 
Daniel Migault
Ericsson