Op 26 okt. 2012, om 02:33 heeft Lorenzo Colitti het volgende geschreven:

> On Fri, Oct 26, 2012 at 1:08 AM, RJ Atkinson <[email protected]> wrote:
> I'm not comfortable with overloading a routing
> protocol for use as a DNS transport mechanism.
> 
> Sorry, I didn't mean "routing protocol". I meant "Homenet DNS Information 
> Distribution Protocol" (HIDP). I would expect a useful way to carry HIDP 
> would be using the homenet routing protocol which the homenet needs to 
> implement anyway, but other implementations such as global IPv6 multicast, 
> its very own DHCPv6 option, SMTP between homenet devices, or carrier pigeon 
> are possible and encouraged. After all, this is the IETF: there's always more 
> than one way of doing things, and at least one of them is always a DHCPv6 
> option.
> 
> But seriously: why are you not comfortable with this idea? We need a routing 
> protocol for the homenet anyway. A link-state routing protocol can carry 
> multiple TLVs, including TLVs for DNS servers. Routing protocols can be 
> authenticated. The devices that need to propagage DNS are likely going to be 
> home routers. Why not use the routing protocol?
Because we need this info in all nodes. I'm not sure we shall have interactions 
between hosts and the routing protocol.

>  
> I'm also nervous about both DNS authorisation
> and DNS authentication.  Who is allowed to make
> which DNS advertisements and how do I authenticate
> the received DNS advertisement as both valid and
> authorised ?
> 
> I don't see a difference. There is no authorization or authentication today. 
> When you get a DNS server via DHCP, you believe it, or choose not to believe 
> it, based on no information at all. If there's a rogue DHCP server on the 
> link that hands you a rogue DNS server, then guess what, you lose. The only 
> thing that would change here is that you would use HIDP to distribute that 
> information more than one hop away. Whatever mechanism that you want to use 
> to authorize and authenticate DNS servers can be used regardless of whether 
> you learn them via HIDP or via DHCP.
>  
>   (NB: With ordinary DNS, the answer is DNSsec.
>    With mDNS, DNSsec also probably can work.)
> 
> Surely there is some alternative approach that
> doesn't require such overloading and complexity.
> 
> Well, let's see. You have an ISP that hands you a DNS server using DHCPv6. 
> You're also connected to a walled garden that hands you a global but 
> partitioned IPv6 address that can only reach the walled garden, and gives you 
> its own DNS server. You want things to work more than one hop away. How would 
> you implement this?
BRDP.

Teco.

> _______________________________________________
> homenet mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/homenet

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

Reply via email to