Russell Nelson <[EMAIL PROTECTED]> wrote:
|Various people have pointed out the difficulty, or at
|least potential difficulties, or arranging for a wireless community
|network to have a connection to the Internet.
Assuming I was one of those people, I meant only to say that there would
likely be friction from ISPs if you try to do it for free. However, see
bottom...
|One possibility which
|people haven't seemed to think about is one of peering with an
|Internet service provider rather than connecting.
Hmm, well, I've thought about it, but it's the basic chicken/egg problem.
It won't happen until the wireless community network is very big.
|We've all gotten so used to having to buy Internet access that none of
|us are thinking about the possibility of peering with a provider.
It is amazing how acclimated the market has become to the concept of ISP
as middle man, getting paid by both sides of the game. But then, I guess
this goes along with accepting the retcon of how (and when) the commercial
internet transition came about. :)
|That is, exchanging packets for free. Peering isn't done so very much
|anymore because the backbone providers have gotten so big that they
|don't *need* to peer with anybody but themselves anymore.
They do still seem to call it peering even though the small network pays
the large one.
|In fact,
|the peering agreements are hush-hush proprietary.
I haven't seen one recently, but historically this was because they tended
to contain terms that might be viewed as anti-competitive. For example,
they might say something like, "we will carry your customers' routes if
you carry our customers' routes and you insure that your customers do not
offer service superior to 9600baud SLIP to anyone else." (The SLIP reference
should date my information. :) I don't think things like this matter much
to the big guys anymore (exactly because they are the big guys) but it could
be a problem if they were worried about protecting their customers' markets.
|But imagine the case where people run popular servers out of their
|homes over this wireless network. One of these servers is *so*
|popular that it serves as a driver of traffic -- e.g. a Slashdot. It
|causes other people to want faster links just so they can get access
|to this server. This is something that other providers want to have
|connecting to their network. Such is the stuff of which peering
|agreements are made.
I have it on reasonably good authority that the only general class of
servers that generates this level of interest is, umm, shall we say,
"picture providers". (In fact, I heard that an ISP bought such a provider
to get its ASN.) It's tricky to come up with a special case like Slashdot
on demand. And you do have to generate the interest on a widespread basis
since it doesn't help to peer with just one provider unless you are merely
using that as a supplement to a conventional connection. Also, keep in mind
that even if you got all the top-level ISPs to want to peer with you, there
would still be a non-trivial cost for the physical connection. (Yes, you
can hope that you are so popular that the other ISPs are begging to install
their equipment at _your_ peering POP, but that's unrealistic.)
Now, I still think it may be beneficial to learn from (and possibly copy)
the strategies used by the original proto-ISPs. At the time of the transition
the entities that were to become the ISPs generally set themselves up as
non-profit corporations chartered with finding ways to provide service to their
members on a cost-recovery basis. The NP status made it more palatable all
around for the new entities to inherit the right to decide who could connect to
the existing infrastructure and who could use the resources (mostly government,
and educational machines--remember, there was no web of vendors trying to sell
you the latest widget at that time) available through it. Unfortunately, the
proto-ISPs were not necessarily tasked with finding the _cheapest_ ways to
accomplish their goals and they often decided to use for-profit management
companies to build and run the networks. Eventually the distinction between
the management companies and the NPs blurred, and the rest is history.
So, how about doing something along the same lines, but headed for a
different outcome:
-Form a non-profit tasked with finding new and effective ways to provide
internet service to the public on an _experimental_ basis. The experimental
part makes you less threatening.
-Generate press and try to get the attention of various government agencies
who might provide grants.
-Try to get private funding. I think the FSF demonstrated that this kind of
venture is not impossible from a financial perspective.
-Use whatever funding you can get to buy transit service from one or more
ISPs, being completely above-board and honest about what you are trying to
accomplish.
-Build and deploy the technology. Here again you have an advantage in
appearing non-threatening because many providers believe their own FUD
wrt the non-scalability of any routing model other than the current
aggregation hierarchy hack. I.e., they will be too busy laughing at
you to worry about competition.
-If the technology works you will eventually reach a state where the only
recurring cost is the ISP transit service and the only cost to add a
new user is some minimal one-time hardware purchase. At that point
you can try to hold out with existing funding until the network really
is big enough to start making peering pitches or you can move to a genuine
cost recovery model. Even in the latter case and even if the network still
consists mainly of "clients" the bulk purchasing effect might make it cheap
enough to support continued growth until you reach the magic size and/or
have resources desired by other networks' users.
IMHO, the biggest risk with such an approach (other than its fizzling before
it gets started) is that it might work too well and begin to look like a good
profit center. It's all too easy to start factoring some administrative costs
into cost recovery and to come up with extra ways to spend money which are not
really necessary, but which can be justified in the name of fairness. It is
similarly easy (again in the name of fairness) to introduce value-based fees
and such. Eventually you end up being just another for-profit. I've observed
this process a number of times over the years, though I can't say with 100%
certainty that the intent to go for-profit wasn't there from the beginning in
some/all cases. Maybe what's needed is a GPL-style agreement for network
service...
Dan Lanciani
ddl@danlan.*com
--
general wireless list, a bawug thing <http://www.bawug.org/>
[un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless
