[ih] Some Questions over IPv4 Ownership

[John Day]

But the real point here is that routing email, is fundamentally no 
different than routing packets, except there is no TTL with email. 
All of the same principles hold.


At 7:40 -0400 2010/10/15, Dave CROCKER wrote:
>On 10/15/2010 12:12 AM, Jack Haverty wrote:
>>If "jack at 3kitty.org" is the example of forwarding you have in mind,
>>that's not how it works.  When I change providers, I move my 3kitty.org
>>service from one provider to another,
>...
>>I do have a bunch of xxx at 3kitty.org mailboxes, and they all must move
>>together.
>
>
>You are emulating a version of exactly the service I described 
>towards the end of my note, modulo the extra forwarding hop. It's 
>key feature is that it is independent of ISPs and it does not 
>require their cooperation.
>
>However, as you note, all of the mailboxes must move together: 
>granularity is at the domain name level, not the mailbox level.  To 
>get per-user granularity, you have to encode it in the domain name, 
>given the way email routing works.
>
>The reason you can have "direct" routing, without having to go 
>through a forwarder is that you control the DNS MX record.  In 
>effect that means an MX record per "customer", if not per "mailbox". 
>Again, that's doable today and it is done today.  The challenge is 
>scaling that model up to a mass market.  In effect, it means an MX 
>per user (or maybe per family).  Yuch.
>
>d/
>
>ps.  I figure you're still unhappy we didn't adopt your encoding 
>scheme when we did RFC 733...
>
>
>
>--
>
>   Dave Crocker
>   Brandenburg InternetWorking
>   bbiw.net