Jakob Schlyter wrote:
>On Wed, 4 Dec 2002, Carl Moberg wrote:
>>>But why would you then need to let the users login since you already
>>>know from what phoneline the user comes from...
>> Statistically, there are more than one physical person behind every
>> phoneline (understatement). It would be extremely bad design to
>> identify the end-user by the way he/she connects to the network.
>same could apply to pstn phone lines, but noone requires a pin code before
>dialing either. if this is a question about liablity, then perhaps the
>subscriber contract should be changed to reflect reality.
No, it's simply a question of authentication, authorization and
accounting (AAA). The
weak form of AAA used for voice over POTS may be sufficient in that
but my bank sure still want a PIN in order to give me information on my
>are loosing customers because of this. the user want a simple always on
>services, not some almost-like-a-modem services (though pppoe providers
>are even worse).
Well, I partly agree. Some users want a simple always on service, most
don't really care if they have to type their username and password,
not if that means that they may get cheaper access. No user wants to
the full weight of always on.
It would probably be really interesting to see hard figures on how much
actually lost per ADSL-customer and year in Scandinavia. The way to go
beloved ;-) ISPs is not towards more "simple" always-on connections
want them to die off).
Why is it that we don't have a lot more *DSL/Ethernet/Barbwire-to-the-home
operators in Sweden. A not so wild guess is that when you start
beans you realize that you will actually lose money starting with the first
connected customer. "Negative economy of scale" if you will.
But, my guess is as good as yours. Things may turn the other way around, I
just fail to see how.
>the cpe nat-box vendors hack strange code into their boxes to login to the
>service all based on a proprietery webthingy. I really hope pana, the ietf
>working group doing "Protocol for carrying Authentication for Network
>Access", can make something that standardizes the authentication protocol
PANA is one shot at it, yes, there are others solutions already in use
customers (e.g. http://www.weca.net/) that aim and shoot for a subset
PANA problem domain.
This whole questions is not primarily one of protocols, technology, or
it's a question of how to actually build and run a sound business on
using the Internet.