In einer eMail vom 03.07.2008 21:47:51 Westeuropäische Normalzeit schreibt  
[EMAIL PROTECTED]:

On Thu,  Jul 3, 2008 at 3:49 AM,  <[EMAIL PROTECTED]> wrote:
> In  einer eMail vom 02.07.2008 23:43:24 Westeuropäische Normalzeit  schreibt
> [EMAIL PROTECTED]:
>
>> Layer-3 addresses  presently describe two characteristics of the
>> endpoint: its  network location and its identity. In a clean slate
>> environment,  it is not obvious to me that path-selection need know
>> anything  about the identity part; it need only know about one of the
>>  network locations.
>
> Right. I'd call it "learning from the  postman".A letter isn't checked at 
the
> ingress postal office whether  it is deliverable or not (whether the 
receiver
> has or has not moved to  some other place). Instead it is forwarded to the
> egress postal office  without such checking. No attempt is made neither as 
to
> inform,  world-wide, any postal office when someone moves to another
> place.  Every year many new residential areas are going to be developed, but
>  no postman has ever complained about an (increasing) scalability  problem.

Heiner,

I'm not sure that speaks to the question. Let  me borrow your analogy:

Line 1: William Herrin
Line 2: 3005 Crane  Drive
Line 3: Falls Church, VA 22042
Line 4: United States

Your  point is that a post office in Dublin need not know about 3005
Crane Drive.  It need only get the mail to the US. The first post
office in the US need  not know about Crane Drive either; it need only
get the mail to the post  office for 22042. The post office for 22042
does, however, need to know how  to get to 3005 Crane Drive.

Correct


While that's undoubtedly true, all of lines 2 through 4 are used  for
path selection at various stages of the letter's trip. As Noel  said,
they are inherently inseparable from the path selection process..  I
could not, for example, tell the post office in Dublin to deliver  a
letter to "3005 Crane Drive, United States" or "Falls Church  VA,
United States."  It wouldn't reach me.

Correct



-MY- point was that Line 1 need not be there at all. It is  an
identifier which serves no role in the routing. If you get line  1
wrong or leave it off entirely your letter will still get to  me.


And this is precisely MY point, too:-)
Ignore line 1 before the letter hasn't reach the egress post  office.
(maybe we should "invent"  MPLS-2 :-).



Line  1 is valuable for other purposes. I generally round-file letters
addressed  to "Current Resident" and I wouldn't want to accidentally
open someone  else's mail. However, there is no inherent reason that
"William Herrin"  must be a part of the address. It would be just as
functional if I found it  on a second envelope enclosed in the first.

This is important, because  as it turns out, letters address to:

Line 1: William Herrin
Line 2:  6857 Lafayette Park Drive
Line 3: Annandale, VA 22003
Line 4: United  States

will ALSO get to me, albeit a little more slowly. The person  with my
identity is reachable at multiple locations each of which can  be
described in a manner close enough to a hierarchy to be  efficient.
I can imagine similar situations:Routing to the egress node based on lines  
2-4 combined with line 1 ="to all residents there"
 
Heiner



   

Reply via email to