[arin-ppml] Policy Proposal: Extend 16 bit ASN Assignments

[michael.dillon at bt.com]

> At this point all the major router vendors have corrected their
> ASN handling code so that 4 byte AS numbers are not a problem
> unless you want to use one yourself.

And they magically updated the software of all their
devices deployed out there? I thought not.

> If I asked the audience at the meeting if they knew what a
> downstream O2 sensor in a car did I'd get the same sea of blank
> faces.
> 
> The people who need to know are the people requesting ASNs

Sorry, but the world of IP network engineering is far bigger
than the in-crowd that hangs out on NANOG or ARIN lists.
The fact is that many IP network engineering folks maybe only
read Light Reading once a month, or look through the Networkers
program once a year when their budget request to attend the
conference is denied. And a lot of them don't speak English
very well, if at all.

> And the problem is that if ARIN asks everyone requesting an
> AS if they really, really, really, really!!! are sure they
> can use a 4 byte ASN, because it MIGHT NOT WORK with their
> gear, why then requestors are simply going to say "fine,
> then gimmie a 2 byte ASN" and they won't even bother learning
> anything more about it.

It would help ARIN if they had a wiki page, even one on the
IPv6 wiki, that documented the first software release supporting
4-byte ASNs for every known hardware platform including stuff 
like Alcatel 7750, Huwaei, etc.

> The folks who get the 32 bit AS and find out after the fact that
> they have to replace hardware to use it, will hopefully be mad enough
> to call their router vendor and scream at them for not releasing
> a firmware update for their gear.  

History shows that instead, they go back to ARIN and say that these
chocolate ASNs don't work, please give me a vanilla one.

--Michael Dillon