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

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

-----Original Message-----
From: ipv6 [mailto:[e-mail]] On Behalf Of David Farmer

> That is why I suggest noting that the classic IPv4 subnet prefix
> is most similar to the IPv6 on-link prefix, which both can be of
> any length, and both are directly related to routing, by
> determining what is locally reachable.

I see many taking this view, but the subtlety gets lost to me. In IPv4, I'm given, say, five class C blocks to use on a platform. The service which assigned me those five class C blocks assumes the prefix is 24 bits wide, when they route packets to these platforms. Then I create 60 subnets out of these five class C blocks, and what I consider "prefix" becomes 26, 28, or other width. How is this different from the way "subnet prefix" is applied to IPv6?

> So, I think we have been arguing about whether the equivalent of the
> IPv4 subnet prefix for IPv6 is fixed at /64 or not, and by tying the
> term subnet prefix to the definition of IID in RFC4291 and its
> predecessors we created an unnecessary conflict. Because the purpose
> of the IPv4 subnet prefix is more closely related to the IPv6 on-link
> prefix not the IID.

Even assuming the fixed length IID, I'm not terribly taken with this argument. If the IPv6 ISP assigns me a /56, that ISP's definition of "prefix" will be 56 bits. When I use that /56 block, my definition of "prefix" will become /64. Not all that different from IPv4, as described above?

Bert