[arin-ppml] The non-deployment of IPv6

[Steve Bertrand]

michael.dillon at bt.com wrote:
>> Although I'm not in a position to request PI space, where I 
>> had a hard time was informing ARIN that a /32 was too much 
>> for my needs (and I _still_ received it, due to policy).
> 
> That /32 may have been too much for YOUR needs, but you are
> not the only one whose needs are analysed. In particular, the
> global community of network operators also has needs, and those
> needs are better served by giving you a /32 even if, on the 
> surface of it, the allocation appears too big.
> 
> 1. We can afford to give every operator a /32, not matter
> how small.

Yes, I know that _now_, but I didn't know then. I looked at my measly
/21, threw the /32 into Perl, Googled how many numbers that was, and was
blown away.

It took me some time to realize that v4 mentality had to go out the window.

I was upset when people would express to me "get rid of the v4
thinking", because I was trying to conserve. It takes quite some time
building up experience and knowledge to fully understand what those
people meant...

> 2. By giving everyone a /32 and encouraging them to announce
> a single prefix, we reduce the profile of a single network
> operator to a single entry in the global routing table. There
> are exceptions of course but that is by far the most common
> situation.

Of course. I truly believe in having a very tight routing table. I am
exceptionally fond of Bill Herrin, his ideology, his work, and the
effort he has put into RRG and other areas.

> 3. If we gave smaller operators longer prefixes, many of them
> would come back for additional, non-aggregatable allocations
> as their business grew. By giving out the /32 to everyone, we
> minimize the impact on the global routing table of many 
> extra non-aggregatable allocations.

Again, agreed. I was pleading with ARIN at the time (iirc, on the
telephone) that "oh my God! it's such a waste of space!", but as I said,
now I know ;)

> P.S. You cannot understand IPv6 piecemeal or by comparing
> IPv4 things to the IPv6 equivalent, because there often is
> no equivalent in IPv6. Twisting the meaning of things in
> IPv6 to conform to an IPv4 world view just leads to confusion.
> Try to understand IPv6 as a holistic system of networking
> that has a family resemblance to its ancestor, but also 
> incorporates new genetic material.

Indeed.

Thanks Michael,

Steve