Jon Postel Re: 202210301538.AYC

Mon Nov 7 12:46:45 UTC 2022

On 11/5/22 8:19 AM, Masataka Ohta wrote:
> William Allen Simpson wrote:
> 
>> Something similar happened with IPv6.  Cisco favored a design where only
>> they had the hardware mechanism for high speed forwarding.  So we're
>> stuck with 128-bit addresses and separate ASNs.
> 

> Given that high speed forwarding at that time meant TCAM,
> difference between 128 bit address should mean merely twice
> more TCAM capacity than 64 bit address.
> 

Carrying the ASN in every packet, going back to my Practical
Internet Protocol Extensions (PIPE) draft that was merged into
SIP->SIPP, meant there was no need for Content Addressable Memory.

And was closer to the original Internet Protocol design of smart
edges with dumber switches.

Reminder, PIPE was 1992.  We'd barely deployed BGP.

> I think the primary motivation for 128 bit was to somehow
> encode NSAP addresses into IPng ones as is exemplified
> by RFC1888. 

Probably as many motivations as there were members of the IESG.
Telcos wanted their addresses, some hardware vendors wanted
IEEE addresses.

But several vendors seemed very intent on using the standards
process for putting competitors out of business.

> Though the motivation does not make any
> engineering sense, IPv6 neither.
> 

Not much about the IPv6 result makes any sense.  I'd reserved v6.
For a long time, I've been rather irritated that it was used for
purposes so far from my design intent.