[sig-policy] COMMENTS REQUIRED ON ASSIGNMENT OF IPv6 ADDRESSES FOR EXCHA
I've wrote a comment on this topic in early Jun. A few additions.
At least one vendor presented that they are going to support BGP
peering w/ link-local addresses. While their current release has some
bug for nexthop field, it is expected to be fixed in the next release.
Still I have no reason why an exchange point do require globally
unique address space. We may need more debate to make clear
technically if we need a separate address to an IX.
Those systems which requires globally unique, routable, and provider
indepenent addresses are only Root DNS servers. Their IPv6 addresses
will be specified in the millions of config files and it takes many
years to most of the config files are updated. The renumbering of the
Root DNS servers would be extremely difficult. More over, if we
allocate separate address space to each of the Root DNS servers, the
routing table growth is only 13 as of writing. This can be grow up by
a few if most of the DNS servers support EDNS0.
If we allow provider independent potable addresses to the internet
exchanges, the number of such IXes in the world could grow
indefinitely without defining very specific criteria.
Renumbering of IX is not so easy, but not impossible.
-- Akira Kato, WIDE Project
* sig-policy: APNIC SIG on resource management policy *
* To unsubscribe: send "unsubscribe" to sig-policy-request at apnic dot net *