> From: Mark Andrews <[EMAIL PROTECTED]>
> EDNS is required to get referrals through which contain
> enough glue in a dual stack IPv4/IPv6 world. DNSSEC will
> just increase the need.
>
> TCP is required to get the larger answers application writers
> are requesting from the DNS.
Agree, but both two have weakpoints each.
EDNS0 weak point is in IP fragmentation. I think EDNS0 can support
4096 or 8192 octet, but 65535 octets is unimaginable.
TCP may not work when BGP route flap occuer or BGP transit network
uses BGP per packet load balancing (which is not allowed in RFC1771
BGP4, but it exists).
How large DNS data can be carried by EDNS0?
- 512 UDP
513 - 4096 UDP with EDNS0
4097 - 65535 TCP?
> Both in my opinion are manditory to support in servers and
> interative resolvers. One an get away w/o EDNS in stub
> resolvers until you require DNSSEC which requires EDNS.
agree.
Transport between authoritative servers and iterative resolvers and
transport between stub resolvers and caching(resolver) servers should
be considered separately. need to consider...
--
Fujiwara, Kazunori JPRS
.
dnsop resources:_____________________________________________________
web user interface: http://darkwing.uoregon.edu/~llynch/dnsop.html
mhonarc archive: http://darkwing.uoregon.edu/~llynch/dnsop/index.html
