Re: draft-kawamura-ipv6-isp-listings review

[Seiichi Kawamura <kawamucho@mesh.ad.jp>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Kurt

Lindqvist Kurt Erik wrote:
> On 21 jul 2010, at 04.40, Seiichi Kawamura wrote:
> 
>> This kind of discussion is exactly what I wanted to have at Maastricht.
>> To what extent, would and ISP need to cover for their service to be
>> called a "working IPv6 ISP"? SMTP? DNS caches? Would a single home
>> tunnel transited ISP be equally considered IPv6?  For example,
>> some ISPs can choose to provide SMTP via IPv4 only but their MX must be
>> dual stacked for their users to communicate with the global internet.
> 
> Personally I believe the criteria for an ISP to be "v6" is that they have a PA block and that they are giving out customer assignments. I would prefer a requirement of native (i.e dual-stack or v6 only) upstream connection. 

Although I'm mostly in agreement with you, having a PA and
assigning it to customers is more the definition of an LIR,
and if I am not mistaken, an ISP is not always an LIR.

One of the cases is where the ISP buys a wholesale ISP service
from another ISP and provides it to their customers. The job
of the non-LIR ISP is to provide customer support, mail services,
CPE delivery,etc and the LIR ISP providing the wholsale does most
of the network providing. I know a few "ISPs" in Japan that
provide internet services like this.

In this case, the requirement for cusotomer assingment of the PA block
falls into the responsibility of the LIR ISP.

So the PA assingning requirement doesn't have to be fullfilled by
the ISP that the customer interacts with, as long as the LIR ISP
involved in the service does so.

Regards,
Seiichi


> Best regards,
> 
> - kurtis -
> 
> 
> 
> 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)

iEYEARECAAYFAkxKj9QACgkQcrhTYfxyMkJ9kgCeLcHeCdJAygQkZHzDYfVAPzux
9kAAn1xX+dovJQbFK029E5iUZ4iJANdX
=bjxh
-----END PGP SIGNATURE-----