Re: [77attendees] Bar BOF: Impact of NAT444 on content providers

["Dan Wing" <[e-mail]>]

The title is misleading.  "NAT444" is the colloquial term for 
"double NAT":  in-home NAT and service provider NAT.  However,
the same problem occurs without the in-home NAT (that is, 
just the service provider NAT).  The problem for content 
providers are large NATs and IPv4 address sharing among 
subscribers in different physical locations, not double NAT.

Going beyond the title to Lorenzo's specific request, this seems 
a grey line to request a single document describing only 
content provider impact.  For examples:

* if my neighbor is running a POST script on a website (e.g., to 'vote
  early, vote often'), and the content provider blocks the IP
  address...  is that something that "impacts the content provider"?
  Seems answer is No.  Or does it impact the other users sharing that 
  same IPv4 address?  Seems answer is Yes.  Same for email spam
  blacklisting.
* If the web application consumes more TCP ports than available on the
  NAT, does that affect the content provider (answer: yes, has to
  design application to deal with that case, or expect application to
  break) or the user (answer: yes - content doesn't display correctly
  or maybe displays slower).
* Does inability to geo-locate affect the content provider (can't
  display targeted advertising) or the user (can't locate nearby
  businesses).

Those are just examples.  I expect we can come up with more
grey areas.

A different layout of draft-ford-shared-addressing-issues might
go a long way towards addressing the desire for a citable 
document on the negative affect of IPv4 address sharing to a 
content provider.  Failing that, perhaps a separate draft; however
it seems difficult to describe any of these as solely or even
primarily impacting the content provider -- it is the collateral
damage caused by the negative reputation of the IPv4 identifier
that is the foundation of much of the woe (for the penalty boxes).

For location delivery *solutions* with IPv4 address sharing -- which I
suppose are out of scope of this Bar BoF, but I believe are why this
document is desired -- I am aware of draft-vandergaast-edns-client-ip
which suggests the first 24 bits of the client's IPv4 address be
included in the DNS query and I imagine (but do not know) that draft
is part of the motivation for this Bar BoF on IPv4 address sharing.
If location is the primary driver, have other ideas been floated and
discussed, such a The-Son-Of-Ident to query a carrier's NAT for a
subscriber's location or query the carrier's NAT for the first 24 bits
of the client's IP address?

-d

> -----Original Message-----
> From: [e-mail] 
> [mailto:[e-mail]] On Behalf Of Brian E Carpenter
> Sent: Sunday, March 14, 2010 4:27 PM
> To: [e-mail]
> Subject: [77attendees] Bar BOF: Impact of NAT444 on content providers
> 
> Bar BOF: Impact of NAT444 on content providers
> 
> Day:  Tuesday March 23
> Time: 17:30
> Room: TBD
> 
> Inspired by some remarks of Lorenzo Colitti (Google) at the recent
> APRICOT/APNIC meeting in Kuala Lumpur, we plan to discuss the value
> of a short draft aimed *specifically* at content providers, 
> to describe
> the impact on them and their customers if many of those customers are
> trapped behind double IPv4 NAT (NAT444). Clearly the issues are not
> news, but they tend to be described in complex general drafts, or
> mixed in with proposed solutions or alternatives to double NAT.
> 
> Quoting Lorenzo: "...it would be
> good to put together an IETF draft that very concisely lists 
> the problems of
> NAT444 from the content provider perspective. For example: 
> worse geolocation
> for targeted advertising and streaming content restrictions; 
> higher latency;
> IP blocking for abuse/spam causing collateral damage, etc."
> 
> (Also see 
> http://www.isoc.org/educpillar/resources/docs/ipv6_200905.pdf)
> 
> The bar BOF agenda is to decide whether to proceed with such 
> a draft and
> if so, to identify the people who will write it and the target date.
> 
> Highly relevant existing documents include:
> 
>  draft-ford-shared-addressing-issues-02.txt
>  draft-azinger-additional-private-ipv4-space-issues-03.txt
> 
> Drafts that touch on the topic in one way or another include:
> 
>  draft-nishitani-cgn-04.txt
>  draft-shirasaki-nat444-01.txt
>  draft-shirasaki-nat444-isp-shared-addr-03.txt
>  draft-shirasaki-isp-shared-addr-04.txt
>  draft-davies-reusable-ipv4-address-block-00.txt
>  draft-boucadair-port-range-02.txt
>  draft-ymbk-aplusp-05.txt
>  draft-thaler-port-restricted-ip-issues-00.txt
> 
>      Brian Carpenter
>      (whose only commitment is to make the bar BOF happen)
> _______________________________________________
> 77attendees mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/77attendees
>