IPv6 Address allocation

Magnus Ahltorp map@stacken.kth.se
16 Jul 1998 01:35:05 +0200

> Now my question .... 15% of 4 billion times 4 billion is an awfull
> lot of IP address space which has been allocated ...  for what
> reasons has these allocations taken place and how will they be used

It is not only a matter of the number of addresses that can be in
use. A lot of addresses has to be wasted to make routing
efficient. For example, it is impractical to route at something other
than bit boundaries, i.e. the significant part of the address (at a
specific router) is the n left bits of the address (indicated with /n
after the address).

To make the routing feasible, routing must also be hierarchical. This
means that there is a top level routing and routing levels beneath
that. In the 6bone, the top level routing is performed by the backbone
routers as you can see on the various diagrams on the 6bone web site.

Hierarchical routing waste even more addresses, since every routing
level must have well defined bit boundaries, or to put it in another
way, the address is divided into smaller parts. If you look in the
Internet draft draft-ietf-ipngwg-addr-arch-v2-06.txt, their suggestion
is to have 3 bits for format prefix, 13 bits for TLA (Top-level
aggregation identifier), 8 bits reserved, 24 bits NLA (Next-level
aggregation identifier), 16 bits for intra-site routing and 64 bits
for an interface identifier.

This adds up to 128 bits.