> kre wrote:
> (Note: I think I'm one of comparatively few people who actually
> support site-id's in SL addresses).

If I understand correctly, what you are advocating for is SL addresses
that are globally unique (by using a mechanism such as embedding the ASN
in the higher bits). Therefore, when two networks merge, they don't have
to renumber one of them (which is what happens today when two networks
using 192.168.0.0/16 merge).

There is one thing that worries me about such a scheme: it does in fact
give a PI address to everybody. It is to be feared that people using
these addresses will make arrangements with their upstreams to have
these leaked in the global table. Money talks. It will not be possible
to technically prevent this, as the very purpose of site ids is to
exchange routing info about SL addresses between two sites. In no time,
this scheme could be extended to the entire Internet.

What I don't like with side-ids in SL addresses is not that that they
give a unique address to everyone, which is great; but that there is no
control over how far the routing of these can go except "don't do it".

I support having no site-ids in SL addresses. The fact that fec0::1 is
pretty much like 192.168.1.1, called re-usability I think, is the only
guarantee that SL addresses would not be perverted into unaggregatable
PI.

As I mentioned before, PA addresses alone are not a satisfactory
solution. There is a need for some independent, globally unique
addresses. I think that these addresses must be subject to a specific
protocol, not a free-for-all ride.

Michel.


--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to