RE: draft-bourbaki-6man-classless-ipv6-00

["Manfredi, Albert E" <[e-mail]>]

-----Original Message-----
From: Brian E Carpenter [mailto:[e-mail]] 

>> I think there are practical alternatives, so I too would suggest
>> not to state flatly that SLAAC requires fixed length IIDs.
>> Anything that requires RAs to work can make adjustments to its
>> IID, in real time.
>
> It cannot adjust its IID length when assigning itself a
> link-local address *before* receiving any RA/PIO messages.

True. Clearly.

> But whatever length N of IID it uses to generate its LL
> address must match the the prefix length 128-N in a later
> RA/PIO with A=1. Otherwise, SLAAC fails. (see Section 5.5.3
> bullet d of RFC4862, as Jinmei-san kindly explained to me.)

Section 5.5.3 states that the prefix length of a valid prefix must be 128-N bits wide, where N is the number of bits of the IID. Then it states that it's the responsibility of the system admin to ensure that prefix lengths are consistent with the IIDs for that link type.

Now, what prevents a link type from having multiple available IID lengths, ready to go, to match varying prefix lengths that might be available? The basic rules are the same. The only difference is that we are no longer held to just one length of IID (for the link type). So it becomes also the host's responsibility, to use an IID length consistent with the length of prefix offered. 

> So in fact the only thing that works is if the equipment
> assumes the same IID length as the router announces (as 128-N).
> Therefore, N must be predefined.

That's the way it works now, but it's hardly the only way it can be made to work. Similarly, IPv4 did not have to remain classful, because that's the way it had been. IP stacks were upgraded, and CIDR became the rule.

>> ULAs and link local would need to create IIDs with no
>> knowledge of the outside world. But not SLAAC.
>
> ULAs have nothing to do with it; SLAAC doesn't care whether a
> prefix is ULA or globally reachable.

I think you missed my point. When the host forms either a ULA or a LLA, the host must do so with no knowledge of anything external to itself. So sure, the IID in those two examples has to be predictable. Instead, there is no valid reason for the IID used in SLAAC to have to be formed independently that way, other than "that's how it's done now."

Once we are freed from the IPX-legacy notion of the IID being formed from the hardware address of the interface, all these preconceived restrictions, derived from that deprecated notion, can be eliminated. Variable length IIDs are ideal for a system with variable length prefixes. Why should we deprecate IPv6 to something less flexible that it can be?

Bert