[address-policy-wg] Policy proposal: #gamma IPv6 Initial Allocation Criteria

> My personal opinion (after far too much wine) is that the routers will 
> probably be able to cope with anything that is thrown at them (again, I 
don't 
> have the facts to back this statement up nor the knowledge to work it 
out - 
> so don't ask for them Randy, other people know far more about what's on 
the 
> horizon that I could ever hope to).

I happen to think that you are wrong. There are real technical
issues with routers that are hard to solve. The first and most
obvious is that as the size of the table gets larger, it requires
more and more computing resources in the router and more bandwidth
to announce/withdraw routes. It's not just a question of RAM sizes
but also CPU power, circuit capacity, and the time required to
process stuff. As we converge more time-sensitive applications 
onto the network, the time delays introduced by huge global
routing tables are less acceptable than they were in 1995.
The second issue is that all of us have lived through the time
when Moore's law, and related laws, have caused electronic
devices to get better, cheaper and faster every year. But we
are now getting to the point where real physical limits are
being reached, i.e. you cannot make circuits thinner than one
molecule. Nobody knows the exact implications of this but
we can be sure that sometime in this century, there will
no longer be any increase in memory capacities, processor 
speed or circuit capacities. It is wrong to make policies
based on the assumption that any problems with routers will
be solvable just as they were in the 80's and 90's.

> Throwing out conservation just because the v6 address range is so vast 
is 
> plain stupidity - the past, in my mind, exists for one reason, so that 
we can 
> learn from it (I might not, but that's a personal problem or at least 
SWMBO 
> tells me so). ISP's might need a /32 (doubtful in most cases) but 
enterprises 
> DO NOT. Yes, enterprises need to multihome (same as ISP's do) but they 
do not 
> need nor will they ever need a /32 (or even a /48 for that matter). How 
many 
> enterprises currently use a class 'A' in current v4 space - and I mean 
really 
> use - I think that question answers itself.

Again, learning from history does not mean doing the same thing
as was done before. An IPv6 /32 address block is vastly smaller
than an IPv4 class A block because it represents a much smaller
percentage of the total address space. Therefore, we have learned
from history and are not making the same mistake. The other historical
lesson that keeps being brought up is "the swamp" although the people
who mention this almost never define what the swamp is or why it
was bad. In my opinion, the swamp was a set of PI allocations of
varying sizes all mixed together in such a way that it was not
possible to aggregate them in routing announcements. It was bad
because it meant that companies would have to announce many small
disjoint blocks, thus consuming global routing table space. In IPv6
this has been fixed. Everybody gets a single /32 PI block and 99%
of those LIRs will never in a hundred years grow beyond that allocation.
Except for those IPv6 /48 microallocations. Are they really such
a good idea? Didn't we learn from the swamp? 

--Michael Dillon