Re: PPPoA server setup how to?

[James Carlson <carlsonj@workingcode.com>]

James Chapman wrote:
> arun b wrote:
>> Again to clear the  confusion....
>> yes with above remote devices I have tested  for PPPoE on ATM
>> can i call this test as PPPoEoA  !!!!!
> 
> Perhaps you mean PPPoE over an RFC1483 bridged encap ATM link? And now
> you're trying to get PPP working over a VC-muxed ATM connection.

You beat me to it.

It's important to note that these are *completely* different animals.
In the PPPoE with ATM case, the ATM link is really just bridging
Ethernet.  It doesn't really "know" that there's PPP traffic inside, as
it only "sees" Ethernet.

In fact, you can carry *normal* IP traffic over that same sort of
bridged ATM link without bothering to use PPPoE at all; some ADSL
providers in fact did just that -- though most were bitten by the PPPoE
bug, some were not.  It's even easier to do this, provides better
performance, and avoids the MTU hassles of PPPoE.

In the PPPoA case, the ATM link is carrying PPP frames.  The VC is
normally dedicated to just PPP in that instance.  There's no Ethernet
involved and thus no PPPoE.  (And a bit lower overhead than any of the
above schemes.)

Don't confuse the two of them.  Except for potentially using the same
raw hardware, they're quite different in terms of design.  For instance,
the ATM termination equipment in the bridged case is a "simple" Ethernet
bridge, and not much else.  In the PPPoA case, it _must_ terminate ATM
and PPP, and then, if there are any other nodes that use the link, it
must act as a router (or NAT) for the traffic that goes over it.  With
PPPoE, those parts are in the end station that runs PPPoE and PPP.

And, no, I wouldn't call the test "PPPoEoA," though that's more or less
what you'd see on the ATM link itself.  The fact that PPPoE is
"normally" run over Ethernet that's then bridged over ATM is more a
monument to Babel than anything else.  ;-}

-- 
James Carlson         42.703N 71.076W         <carlsonj@workingcode.com>