[address-policy-wg] Easy to remember IP-address

Garry Glendown wrote:

>>Anyway, 4 byte addresses of IPv4 is not so bad to memorize.
>>
>>You are right, however, that IPv6 addresses are impossible to
>>remember not only for end users but also for network operators,
>>which is one of a reason why IPv6 is NOT deployable.

> Sorry, but you must be one sieve-brain (is there such a word in
> english?) of a network operator if you aren't able to remember your /32
> prefix ..

So? Note that there is no guarantee that IPv6 prefix length is /32.

The fact is that many, if not all, network operators can't remember
IPv6 addresses, which is a major, among many, obstacle against
IPv6.

> Please stop spreading FUD about IPv6, just because _YOU_ are either
> incapable or unwilling to deploy IPv6 doesn't mean that the other
> currently >2000 prefix owners would prefer to wait another n years to
> see whether your RFC-draft would be implemented by one single vendor ...

2000 prefixes in some private network 15 years after rfc1883???

No, my interest is on >300000 prefixes routed over the real internet.

> Let's see. "ipv6.google.com". Looks easy to remember to me.

I'm not talking about end users.

>>Instead, everyone should be able to remember 4 byte IPv4
>>addresses and 2 byte short port numbers as was documented in
>><draft-ohta-e2e-nat-00.txt>

> Exactly. Because "www.site1.de:1234" and "www.site2.com:1324" and
> "www.site3.org:1243" are so d at mn easy to remember for the
> average-joe-enduser ...

Exactly, which is why my draft says:

	assign small port numbers evenly to all the customers.

Read it.

A user may be assined ports 100, 199, 298, ... while another user may
be assigned ports 101, 200, 299, ...

> compared to impossible to remember
> "www.site1.de", "www.site2.com" and "www.site3.org", which unnoticed by
> the enduser are resolved to either IPv4 or IPv6 addresses and just
> simply work ...

Let DNS map from domain names to 6 byte address+port is no more
difficult than to let DNS map from domain names to IPv6 address.

> now tell me, what exactly does the end user gain from
> sticking with YAPOSNL? (yet another POS nat layer)

End to end transparency.

						Masataka Ohta