RE: 6MAN WG Last Call: draft-ietf-6man-rfc3484-revise-05.txt

[Dave Thaler <[e-mail]>]

> -----Original Message-----
> From: Dave Thaler
> Sent: Monday, February 13, 2012 2:01 PM
> To: Dave Thaler; 'Chris Grundemann'; 'Brian E Carpenter'
> Cc: '[e-mail]'; 'Brian Haberman'; 'Bob Hinden'
> Subject: RE: 6MAN WG Last Call: draft-ietf-6man-rfc3484-revise-05.txt
> 
> Yet another problem in draft-ietf-6man-rfc3484-revise...
> 
> Section 2.4 (Private IPv4 address scope):
> [...]
> >   The algorithm currently specified in RFC 3484 is based on the
> >   assumption that a source address with a small scope cannot reach a
> >   destination address with a larger scope.
> [...]
> 
> The above sentence is simply not true, it was NOT based on such an assumption
> at all.  It was based on the assumption that it was
> less likely to work.   There's two reasons why it's less likely to work.
> First, it might or might not be able to reach it (the text overstates by saying it
> cannot... it was acknowledged that it may or may not).
> Second, if it goes through a NAT, it might not work for protocols that embed IP
> addresses in payloads.
> [...]
> 
> >   Due to this assumption, in the presence of both a NATed private IPv4
> >   address and a transitional address (like 6to4 or Teredo), the host
> >   will choose the transitional IPv6 address to access dual-stack peers
> >   [I-D.denis-v6ops-nat-addrsel].  Choosing transitional IPv6
> >   connectivity over native IPv4 connectivity, particularly where the
> >   transitional connectivity is unmanaged, is not considered to be
> >   generally desirable.
> >
> >   This issue can be fixed by changing the address scope of private IPv4
> >   addresses to global.
> 
> Section 10 of RFC 3484 contained many examples.   -revise contains
> no such example of what it's talking about, so I have to guess.  Let's look at 3
> cases.
> 
> Case 1:
>  D set = { global IPv6, global IPv4 }
>  S set = { Teredo IPv6, RFC1918 IPv4 }
> 
> Under RFC 3484 rules, Destination Address Selection would prefer the Teredo
> connectivity under rule 2 (Prefer matching scope).
> 
> Under -revise rules, Destination Address Selection would still prefer the Teredo
> connectivity under rule 6 (Prefer higher precedence), since the precedence of
> the (non-Teredo) destination address
> beats the precedence of the IPv4 address.   Hence -revise
> does not change the behavior in this case.

Dmitry Anipko pointed out that rule 5 (Prefer matching label) would cause
the -revise rules to prefer IPv4.  Still, I'd prefer a solution that doesn't solve
this problem by creating another one (case 3).   That is, we should fix a problem
rather than just move it around.

I'll think about this and  see if I can come back with a proposal.

-Dave

> 
> Case 2:
>  D set = { Teredo IPv6, global IPv4 }
> 
> Not an interesting case because Teredo addressing should be disabled when a
> host has a global IPv4 address.
> 
> Case 3:
>  D set = { global IPv4 = 1.2.3.4 }
>  S set = { NAT-ed IPv4 = 10.2.3.4, global IPv4 = 128.66.3.4 }
> 
> Under RFC 3484 rules, Source Address Selection would prefer the global IPv4
> address under Rule 2(Prefer appropriate scope).
> Under -revise rules, Source Address Selection would instead prefer the NAT'ed
> IPv4 under Rule 8 (Longest matching prefix).
> 
> This is broken.   I don't see a real case the proposed change
> fixes, I only see real cases it breaks.
> 
> -Dave