[e2e] on local ethernet throughput?

[Woojune Kim]

While I agree that it was an incredible kludge that was almost forced by
the carrier's "interest" in ATM, I think there may have been other
factors. 

For example, I've used DSLs in Korea. (Which by the way has over 4
million DSL lines - the largest in the world) The local operator offered
two modes of connecting. 

One was by using a USB connected DSL modem and the other was an ethernet
connected DSL modem. 

For the first case using PPP was natural as MS (and linux etc.) all came
with simple dial up  software (based on PPP) that could be easily used
for IP transport. 

The second case is where PPPoE was used. 

Using PPPoE in the second case had the advantage of offering
authentication / accounting in a manner similar to the methods already
used for the first case and also the operator's previous dialup modem
connections services. I think for the operator these legacy cases (which
probably involved lots of RADIUS servers and training personal in
debugging PPP related problems) also pushed them in their decisions.

So, I'd submit that even if packet over DSL had been available, many
operators would still have tried the PPPoE path.


-----Original Message-----
From: David P. Reed [mailto:dpreed at reed.com]
Sent: Wednesday, October 24, 2001 4:51 PM
To: Andrew Smith; Woojune Kim; end2end-interest at postel.org
Subject: RE: [e2e] on local ethernet throughput?


The issue as I understand it was that the only termination equipment 
available for companies to terminate DSL at the CO was the so-called
DSLAM 
(invented for ATM circuit switching).  And the customer side equipment
was 
Ethernet on one side and ATM over DSL on the other.

PPPoE only capped the kludgery involved.

If equipment companies could have ignored the ATM madness of bell-shaped

companies (carrier class equals ATM...), then we would have had packets 
over DSL, and an Ethernet or IP-based authentication (like 802.1x).

At 01:23 PM 10/24/2001 -0700, Andrew Smith wrote:
>There are plenty of other possibilities that involve "user"
authentication
>in the control plane that don't need to touch the data plane packet
>encapsulation. One example is what we did for IEEE 802.1X - an on/off
switch
>for most data packets and a way of intercepting a control plane message
(EAP
>over a new L2 encapsulation) were the only data plane intrusions. A
much
>cleaner design, I think (only works for a switched pt-to-pt topology,
or
>some emulation of that like 802.11 wireless, of course).
>
>I put "user" in quotes above because I think that what the DSL
deployers
>wanted was "payer authentication", not necessarily "user" (I have to
give my
>"user" password to anyone in my household for them to use the DSL link
with
>a PPPoE session and my provider makes me use that same password to use
their
>email services :-( (hence the new email address!).
>
>But this isn't really the place to gripe about the power of monopolies
etc.
>even though it impacts the "end-to-end-ness" of the service ...
>
>Andrew Smith
>
>
>-----Original Message-----
>From: end2end-interest-admin at postel.org
>[mailto:end2end-interest-admin at postel.org]On Behalf Of Woojune Kim
>Sent: Wednesday, October 24, 2001 12:29 PM
>To: end2end-interest at postel.org
>Subject: Re: [e2e] on local ethernet throughput?
>
>
>Hello,
>
>I was just wondering about the comment w.r.t PPPoE being an
abomination.
>
>
>While I agree it is a complete mess and not very pretty, I think it was
>deployed because network operators had a real problem : authenticating
>users and making money when deploying DSL modems.
>
>I've always wondered if there are any other methods that could have
been
>used.
>
>thanks
>
>============================================
>
>Date: Wed, 24 Oct 2001 09:11:41 -0600 (MDT)
>From: Vernon Schryver <vjs at calcite.rhyolite.com>
>To: end2end-interest at postel.org
>Subject: Re: [e2e] on local ethernet throughput?
>
> > From: Graham Cope <G.Cope at ftel.co.uk>
>
> > > we still teach students about bridging?
>
> > And some of us in the DSL community might suggest that you teach
them
> > about PPPoE as well.
>
>If people had been taught the facts of bridging instead of merely
>the trade rag and salescritter "bullet items," we might not have
>the abomination that is PPPoE.
>
>
>Vernon Schryver    vjs at rhyolite.com

- David
--------------------------------------------
WWW Page: http://www.reed.com/dpr.html