Mukund Sivaraman wrote:
> Maybe it was not clear from the original mail: when a different
> mechanism (port and related protocol change) is being introduced for
> encrypted traffic which current implementations cannot use without
> changes to them, it would be good to consider revisiting the DNS message
> format.

>From the DPRIVE charter:

    The DNS PRIVate Exchange (DPRIVE) Working Group develops mechanisms
    to provide confidentiality to DNS transactions, to address concerns
    surrounding pervasive monitoring (RFC 7258).

    …

    DPRIVE is chartered to work on mechanisms that add confidentiality to
    the DNS. While it may be tempting to solve other DNS issues while
    adding confidentiality, DPRIVE is not the working group to do this.

DNS transactions are carried over DNS messages, and RFC 1035 §4
"MESSAGES" divides the message protocol into a "Format" (§4.1) and a
"Transport" (§4.2).

DPRIVE has introduced a new "Transport" backwards incompatible with
existing clients (DNS-over-TLS), but it hasn't introduced any backwards
incompatible changes to the message "Format". There have been backwards
compatible changes, though (EDNS0 padding), using the existing DNS
message format extension mechanism, in support of transaction
confidentiality.

If I understand correctly, you'd need to show that making backwards
*in*compatible changes to the DNS message format are necessary to
providing DNS transaction confidentiality. (Or re-charter.)

> The reasons for this have been given, and some of these are
> things we have recently had performance or filtering issues with with in
> the implementation I work on. After a point, no amount of tweaking the
> implementation is going to help an inferior format.

What does poor message decoding performance from a particular
implementation have to do with DNS transaction confidentiality?

(There are DNS implementations that advertise high query rates with
filtering, so I'm skeptical that backwards incompatible changes to the
message format would deliver substantial speedups.)

-- 
Robert Edmonds

_______________________________________________
dns-privacy mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dns-privacy

Reply via email to