On torsdag, dec 12, 2002, at 07:47 Europe/Stockholm, Måns Nilsson wrote:
> Find a killer app that won't work without real Internet.
> IP telephony, yes.
Exactly. I have handed out some 200 SIP-based phones to people "which
should know better" and contrary to what they think, I don't give out
the phones because I am nice. I give them out because they before
getting the phone do not understand how bad it is to have NAT between
an end node and the public Internet.
Now they know.
The Internet Architecture has three layers, while the traditional
architecture have only two:
What people need to understand is that the applications are services
_are_ disconnected from the transmission layer, and that access to IP
has nothing to do with the services.
As Calle says, ISP are trying with whatever they can to glue the
applications to the IP layer. Why? Well, this is simple, they create
bad services and applications.
If they did a good application, they can sell the service to everyone
which have IP access. Not only their own customers. Because of this,
the part of the ISP which sell the service was to have a group of
target customers which is as large as possible. If the service is good,
customers of other ISP's will buy their service.
Now, if they instead create a bad service, customers (even their own IP
customers) will try to buy the application/service from someone else.
And their service will loose customers and not make money.
Instead of letting the customers decide, they create an artificial
market where the customer only can choose between buying the service or
not. Buying the Radio-Shack phone will not be an option.
On top of this we have the fact that services are so darn cheap to
produce, while the IP connection is expensive. I don't think email (as
an example) is possible to do efficient in the span 20-1.000.000
customers. That means that either you have <20 customers or > 1.000.000
customers on your email system. How many inhabitants do we have in
Sweden? 9M Ok, those two rules say that we only need maximum of 9
providers of email, and I think it is as low as 5. This _INCLUDES_
email for enterprises, universities etc.
What I try to say is that we have far too many providers of services,
and that increases the price for the application/service. Instead, the
service is cheap (I think email costs about 100SEK/year and person for
whatever you have been dreaming of) and the IP packets are expensive.
An ISP because of this need to understand they have to make sure they
make money on the movement of IP packets, and then let people which run
services do so. That can be a separate part of the same company of
course, but the keyword is separate.
If they continue to think they can not make money on the IP transport,
they will die. If you're an ISP, just ignore the email, homepage,
whatever service, and instead sell IP packets. Just IP packets. How
many staff can you remove? How many computers from your computer room
can you remove? You can suddenly decide whether you are acting on local
(geographical) or global scale. On global scale you can probably not
have end-user support, but instead only sell to ISP's which act on
local scale. ISP's which have 100-200 customers.
If another corner of your company want to do website hosting, fine. Let
them do it, but see that they make money on their own. Possibly by
giving hosting not only for your IP customers, but also IP customers of
Etc etc etc...