>No. There are plenty of use cases where DNSSEC validation isn't needed or want >ed: armies of dumb IoT devices that have feeble CPUs and die when their batter >ies run flat, stub resolvers using encrypted transports that are considered "g >ood enough", enterprise networks that can't/won't drink the DNSSEC Kool-Aid, e >tc. Foisting (unwelcome) validation that's doomed to fail in these scenarios i >sn't going to have a happy ending or lead to the universal deployment of DNSSE >C. > >You're also redefining the meaning of the DO bit. It was introduced to signal >to a server that the (stub) resolver understood DNSSEC related RRtypes => it w >ouldn't fall over if those RRtypes were included in a response. RFC3225 says n >othing about validation at all. If this matters, I suppose we can always burn >another EDNS header bit to say "I [don't] do DNSSEC validation".
When looking at this from the non-validating resolver that requests something from a domain signed with ML-DSA-44: - the resolver doesn't know that it should expect a large answer. A validating resolver can directly send the query over TCP, but the non-validating resolver has to retry over TCP (unless the non-validating resolver starts tracking how zones are signed). - if the non-validating resolver has no downstream clients that validate then these large signatures take of space in the cache that is just wasted. So a non-validating resolver may have a pretty strong incentive to send queries with DO=0 and only send DO=1 queries when the request it is working on has DO set. Currently that behavior is not allowed, however resolvers do other things as well that are not allowed. So maybe we wait for resolvers to feel the need and to start experimenting with DO=0 queries before we update the DNSSEC standard. I highly doubt that resolvers will complicate their caches just to please authoritative servers. _______________________________________________ DNSOP mailing list -- [email protected] To unsubscribe send an email to [email protected]
