Re: Updated IID length text

[Fernando Gont <[e-mail]>]

On 01/19/2017 12:00 AM, Lorenzo Colitti wrote:
> On Thu, Jan 19, 2017 at 9:37 AM, Brian E Carpenter
> <[e-mail] <mailto:[e-mail]>> wrote:
> 
>     However, consistent use of Stateless Address Autoconfiguration
>        (SLAAC)[RFC4862] requires that all interfaces on a link use the
>     same length
>        of Interface ID. To guarantee interoperability of SLAAC, a fixed
>     length of
>        Interface ID is necessary.
> 
> 
> I'm not a fan of this text, because it's a weak argument. A possible
> (uninformed) response to it might be "why do we need this SLAAC thing
> anyway? I want to use /120 prefixes and DHCPv6, just like I do in IPv4".
> And really, SLAAC is only one of the reasons why we have a 64-bit IID.

Please enlighten me. The 64-bit length has to do with SLAAC, and the
fact that the IID had to be able to embed 64-bit link-layer addresses.

If you were to start with a clean slate, you wouldn't embed link-layer
addresses in the layer-3 address, and hence wouldn't need to do
fixed-length 64-bit IIDs. With manual configuration or DHCPv6, you would
support "any size that can fit all your systems and allow your network
to grow". And with SLAAC (+ something like RFC7217), it would be the
same, with the contraint that "the size results in low host density, so
that IID collisions are not a concern".



> There are lots of good arguments for this in RFC7421 - solid arguments,
> about address scarcity and future flexibility, and so on. 

That translates well into "well... we have so many bits that... let's
burn them in any possible way" (fwiw, I don't think that's the argument
being made in RFC7421, though)

An artifact of history need not and does not automagically become a
design principle.

Thanks,
-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492