Re: Regionally aggregatable address space for multihoming

[Iljitsch van Beijnum <iljitsch@muada.com>]

On Mon, 11 Jun 2001, Brian E Carpenter wrote:

[Problem with geo addressing: regions are big]

> > When the number of routes per area grows, it is likely that the areas
> > themselves shrink, if only for this very reason. I'm not saying geographic
> > aggregation will magically make all the problems go away. However, it will
> > buy us two or three orders of magnitude, which is nothing to sneeze at.

> We are not communicating. You deleted the text that my initial "this"
> refers to, but I think the exact opposite is true: the areas will have to
> grow, since ISP interconnections are essentially random with respect to
> geography.

I guess this is true for the really large networks. Maybe this is because
they've lacked a good incentive to interconnect at more locations. Geographic
aggregation could be the necessary incentive. They are already present in
many locations and local fiber is cheap, so it could happen. On the other
hand, it may not.

> Also the cost pressure is going to be against local
> exchanges and in favour of large regional exchanges. Given this (and
> other points noted below) I don't see that geo addressing buys us
> anything much at all, certainly not orders of magnitude.

I'm inclined to believe that there will be more local interconnection, but of
course everything has to work regardless of what we believe.

Do you agree that it would buy as at least one order of magnitude? That means
three regional exchanges for North America and Europe and one for every other
continent.

> The only safe design target is 100% of all businesses, i.e. several orders of
> magnitude bigger than today's Internet. I see nothing that would limit it
> to 10% or any other particular level; it is likely that multihoming just becomes
> the normal thing to do.

100% would be something like a billion. I think there is a natural limit
because connecting to two ISPs is twice as expensive as connecting to one.

A billion routes means we absolutely have to aggregate on something, probably
on more than one thing at a time.

> Don't misunderstand me - we need a radical solution. I just don't think that
> geo addressing, which is fundamentally the telephone solution, is radical
> enough.

Hey, I'm just as radical as the next guy. But if we can't come up with a way
to make a billion routes work before there are a million, the routers will
choke and we don't have an internet anymore.

I think I'll write up a draft on introducing geographic information into the
current routing architecture. In BGP this could be a community, in IPv6 it
could also be the top 8 bits in an end user /48. Then every user can decide
for herself to implement it or not.

Maybe it would help to make it possible to aggregate on other things than
just the beginning of the address. IPv6 addresses are large enough to hold
lots of good stuff we can potentially aggregate on. For instance, we could
use two 16 bit field that hold the AS number of the two ISPs for a multihomed
network. Then an ISP would simply announce aggregates like:

4567:ISP::   mask ffff:ffff::
4567:0:ISP:: mask ffff:0:ffff::

With every kind of aggregation there is the problem of holes. If the customer
is disconnected from ISP 1 and ISP 1 still announces the aggregate, ISP 1 has
to know it has to forward the traffic to ISP 2 at some point. With just two
ISPs this can be done without exchanging any more specifics by just dumping
the traffic at the other network if there is no local (= this network) more
specific route to the customer.

Iljitsch