Re: Using DNS for SSM

[ken lindahl <lindahl@ack.Berkeley.EDU>]

On Thu, 15 Feb 2001, Dino Farinacci <dino@procket.com> wrote:
>>> I've just submitted a draft which I think the SSM working group should be
>>> interested in.  It can be found at:
>>> http://www.ietf.org/internet-drafts/draft-schoch-dns-ssm-00.txt
>
>    An even simpler approach would be to have DNS records like:
>
>        radio.example.com  AA     10.1.1.1.232.1.1.1
>
>    then 1) the application has the complete channel address for IGMPv3, and 2)
>    you don't have to require application semantics for a DNS name.
>
>    And we have existence proofs that adding new record types to DNS 
>    (particularly simple ones) is not too much effort.
>
>Dino

this has the distinct advantage of requiring only a single dns query to
to get the information, if you're going to add a new RR type, why not add
one that specifically has two fields in RDATA, one for source IP and one
for group ("channel") IP?

simpler still would be to simply have two A records:
   radio.example.com	A	10.1.1.1
   radio.example.com	A	232.1.1.1
and let the application determine which is source and which is group
based on the address range? (i'm prepared for someone to tell me that
there are resolvers out there that pass only on A record to the app.)

multiple sources is an easy extension: just add more A records with the
sources' IP addresses.

ken