Question regarding RFC1897?

[Francis Dupont]

 In your previous mail you wrote:

   While setting up our internal networks and DNS with RFC1897 addresses,
   I became puzzled by the address format.
   
   Assuming that there are very good reasons to use the *providers* ASN,
   wouldn't it make more sense to use the IPv4 Network Address to represent
   the Network Address given to you by the provider?  In other words
   If you have a class A address the IPv4 Network Address field would be
   0x0000AA, class B address would be 0x00BBBB, and so on.   Then the Subnet 
   field could contain the remaining bits.
   
=> in the current RFC the network number are "big endiens",
ie 0xAA0000, 0xBBBB00, 0xCCCCCC but it doesn't matter.

   As it is currently defined, it is difficult to setup reverse DNS records.
   Providers will have to maintain records for each IPv4 subnet that sites
   want to run with IPv6.  Collisions may occur when sites don't work with
   their providers.  Or in some cases, sites won't setup the reverse DNS just 
   to avoid this problem. 
   
=> I don't see any problem, you have just to work with your provider.
It is already the case for IPv4 networks since CIDR...
Don't forget for the reverse domains any reasonnable BIND version is
enough because they use only PTR RRs.

   Ofcourse, if you use your own ASN, the problems go away. 
   
=> which ASN ? You can have many ASN, the provider ASN is registered
then without ambiguity. You have a better aggregation too (perhaps
not as good as we'd like to get).
Providers will be involved when you'll use the provider-based addressing
scheme then I think it is a good thing to involve them without
real burden today...

Regards

Francis.Dupont@inria.fr