On 2012-03-30 20:35, Michael Richardson wrote:
>>>>>> "Brian" == Brian E Carpenter <Brian> writes:
>     Brian> Front posting: I think we are using "walled garden" to mean
>     Brian> several things and that is confusing.
> 
>     Brian> In my mind it refers to a captive customer scenario where a
>     Brian> service provider is intentionally limiting a customer's
>     Brian> access to the global Internet or (by playing DNS and/or HTTP
>     Brian> proxy tricks) presenting a distorted view of the the global
>     Brian> Internet.
> 
>     Brian> This becomes especially obnoxious if the customer has
>     Brian> multiple providers that attempt to enforce different walled
>     Brian> gardens. But that's a MIF issue, I think.
> 
>     Brian> That is quite different from stating that a customer network
>     Brian> should be separated from the Internet by a security fence of
>     Brian> some kind and may also need a local namespace. I thought that
>     Brian> was normally called an intranet.
> 
> I actually can't see a technical difference.

In an enterprise intranet, it's (supposedly) a fact that all routing,
firewalling, etc. is administered by the same IT management. Also,
multihoming takes place outside the intranet - the "walled garden"
boundary is also the multihoming boundary.

In a customer/provider scenario, there is split management, and furthermore
the multihoming boundary is inside more than one walled garden at the
same time. That's why it's messy, IMHO.

> Most intranets work by:
>      1) intentionally limiting a customer's access to the global Internet
>      2) playing DNS tricks
>      3) HTTP proxy tricks
>      4) having a local name space.
> 
> I think that walled-gardens are the political NAT44 of IPv6.

IPv6 makes more walled garden tricks possible, but both intranets and
walled gardens abound in IPv4.

> 
> We can, if we want, stick our head in the sand like we did for NAT44,
> letting the marketplace produce create an arms race between network
> operators and application creators.  Or, we can do like we finally did
> with BEHAVE and specify something.

Good luck balancing on that third rail ;-)

   Brian

> 
> I'm here in homenet because I care about e2e, I want to avoid NAT66, and
> I see great utility in having globably unique addresses everywhere, even
> if they are not globally routable/accessible.
> 
> (ps: I shipped NAT44 products in 1994. I hate myself for helping kill
> e2e. I shipped it with split-horizon DNS standard.  It was all stupid. )
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> homenet mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/homenet
_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet

Reply via email to