[ppml] Policy Proposal 2005-9: 4-Byte AS Number

[Rich Emmings]

On Wed, 21 Dec 2005, Owen DeLong wrote:
>
>>
>> IP addresses in aggregate express a hierarchy.  AS numbers do not.
>>
> Or at least do not yet.
>

I'll hope not, pending a *real* good reason.

>
>> 2 byte ASN's 0 - 65535 unsigned, and 4 byte are 65536-4294967295.  ASN
>> 1,000,000 would mean the AS space has grown an additional ~15 times the
>> existing space.
>>
> Not quite.  16 bit ASNs are 0-65535 unsigned.  32 bit ASNs are 0-4294967295
> also unsigned.  32 bit ASNs which overlap the same integer space as the
> 16 bit ASNs are equivalent and all 32 bit ASN software is required to
> support 16 bit representations thereof.
>

Point conceded: the low range can be 16 or 32 bit, and the high only 32. 
When all systems are 32 bit, then it doesn't matter.

(Problem is if a 32 Bit number gets stuffed into a 16 bit data bucket and 
does something strange, but that's a problem for the routing code 
programmers to not screw up.)

>
> How many 7 digit phone numbers have you seen without separators?
> Most I've seen are either XX-XX-XXX or XXX-XXXX.
>

I'll wager when you recall a phone number, you just rattle off all 7 digits, 
and don't recall or vocalize the delimiters that are commonly used in 
writing.  If it's written down, I'd guess that old fashion , or . delimiters 
depending on the side of the ocean you're on, will work for human representation.