ietf
[Top] [All Lists]

Re: What exactly is an internet (service) provider?

2004-06-22 16:08:40
On Tue, 22 Jun 2004 11:11:25 +0900, Masataka Ohta said:
With your motivation explained and with your three new categories,
all of which are unrelated to telecommunication providers but
related to hotels, I think I can understand your fundamental
mistake.

More generally, Internet access is of 4 basic time-base categories:

1) Access at the location(s) you usually are (home, office, dorm, etc) - these
are "traditional" ISPs.  The timeframe for (business) associations here is 
"weeks to years".

2) Access when you're semi-stationary while not where you usually are (hotels,
extended stays with friends/relatives, and so on).  Timeframe for associations
is "one to several days".  Note *carefully* that this has some corner-case
implications - if I'm visiting my brother in New England, I'm a semi-long-term
(a week or so) transient on his home Ethernet and thence to his cablemodem
provider, while *he* is a long-term user of the *exact same connection*. As a
result, *his* expectations and *mine* regarding the same service may be quite
different (for example, he may be quite OK with the concept that all SMTP gets
redirected to the provider because he uses their mail relay by default anyhow,
but that may be totally unacceptable to me)....

3) Very short term access while stationary long enough to consume 1 beverage -
the kiosk/cafe model.  Associations on the order of 15 minutes to a very small
number of hours.

4) Roaming access while *not* stationary - citywide wifi networks and the like,
where the timescales are seconds to minutes... 

Note that the types of services provided is generally orthogonal to the 
timescale
of the business model - a provider can (for instance) provide WebTV-style 
captive
access at all 4 timescales.  As a result, we probably end up with a 2-D 
matrix...

That is, your draft should have been titled:

      Terminologies on Telecommunication Service Categories of
      Accomodation Service Providers

Given that your stay is temporary, permanent addresses are of
little value. Moreover, as most, if not all, hotels use NAT,
it is of little value to mention global addresses.

Note that the fact that I'm there for only several days does *not* imply that
permanent (at least for the duration of the stay) addresses are of little
value. Also, note that there's a distinction between "permanent" and "routable"
in this context.  There's a wide range of things where I don't *care* what my
current IP address is as long as it's routable - in those cases, I can easily
deal with a DHCP-assigned routable address, but not with a NAT'ed address,
whether the NAT'ed address is dynamic or static.  I can be 192.168.10.10
from last year till the next Winter Olympics - that address Simply Will Not
Work for many things.

In particular, the fact that most hotels use NAT is quite likely one of the big
stumbling blocks for commercial deployment of any application protocol that has
trouble playing nice with NAT (See Keith Moore's list - all of those are
basically not doable in your organization if you have frequent road warriors
who might be participating from a hotel room).

So, in the new draft, you should avoid the word "ISP". Instead,
you can, for example, say ASAP (Application Service Access
Provider).

Providing a common list of such things is the intent of the draft, I believe.

Mobile operators in Japan does provide access to web services,
though not over IP, that they are WSAP. Or, do you like better
calling them Web Access Service Providers?

Oh my... A can of worms indeed.  John - it's your call if you want to
expand the scope to include that class of access or not....

Attachment: pgpvpasqFkqSZ.pgp
Description: PGP signature

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf
<Prev in Thread] Current Thread [Next in Thread>