[12649] in Commercialization & Privatization of the Internet
Re: The CMU Internet Billing Service
daemon@ATHENA.MIT.EDU (Paul Robinson)
Wed May 25 09:11:04 1994
Date: Wed, 25 May 1994 04:14:53 -0400 (EDT)
From: Paul Robinson <tdarcos@access.digex.net>
To: Bernie Cosell <bernie@fantasyfarm.com>
Cc: Everyone Else Lurking on Com-Priv <com-priv@psi.com>
In-Reply-To: <5987:73537@fantasyfarm.com>
On Tue, 24 May 1994, Bernie Cosell wrote:
> In <Pine.3.89.9405231642.A488-0100000@access3.digex.net> on May 23, Paul Robinson wrote:
>
> } Bernie Cosell (bernie@fantasyfarm.com) wrote:
> }
> } : Moreover, I think that the future of the net is clearly in
> } : 'commercial' services --- most folk will be more
> } : interested in the network being *useful*, rather than just being a
> } : source of stuff-to-read.
> }
> } Someone want to explain what 'the CMU Internet Billing Service' is?
>
> You've jumped in about 20 or 30 posts down in the thread --- it'd probably
> be best to try to find the original stuff rather than rely on folks
> biased/incorrect/mistaken summaries of what CMU proposed.
>
> } ...But let me state my own results as
> } suggesting (in alt.games.doom) that there be FTP sites that charge
> } minimally for access so that they don't have to throttle connections.
>
> This isn't the point I made right there [at least not the one in the
> bit of my post you chose to cite right before this remark]. The
> fact is that the spectrum of network services available today is
> just not of much use or interest to most 'ordinary' folk. Try asking
> the guy who runs the hardware store, or your insurance agent, or the
> guy who repairs your car, how handy they would find ANY Internet
> services you can identify to them. We techies find it useful
> [it is nice to be able to get Adobe/microsoft/etc materials 'from the
> horse's mouth]] and folk who enjoy type-type-typing rather than
> actually conversing with people in-the-flesh will tolerate
> the astoundingly prolix and low s/n discussions on usenet. Ask your
> non-computer friends [say, a local carpenter] whether they'd rather
> read something nicely printed, sitting in their easy chair, or sitting
> at a desk staring at a CRT for hours and a time. It is nice that
> I can get Dave Barry's column over the net, but (1) I usually have read
> it [in the local paper] before it arrives here, and (2) I find
> reading the paper {where I can take it to the John with me, or out
> on the back porch] a LOT nicer than reading it on-line.
>
> } : Right now those happen via off-line billing but it would REALLY
> } : simplify matters if you could do over-the-net billing [consider
> } : being able to order a book by emailing the publisher with some
> } : secure billing-mechanism [emailing your mastercard number is
> } : considered fairly foolhardy] -- would that be bad??; or to be able
> } : to subscribe to an on-line [edited, well done, etc] magazine easily
> } : and over-the-net [instead of, again, having to go out of band to
> } : arrange payment]
> }
> } Some people think e-mailing credit card numbers is foolish. I think the
> } chance of the text of a message being captured are fairly low, since
> } so many messages go out and you would have to be on the initial or
> } intervening sites and be watching for traffic.
>
> I guess you missed the big netcom scandal recently. They managed
> to steal *thousands* of site/user/password combinations by simply
> watching the traffic that whizzed by.
Oh no! I read some important material about it. I run a security issues
mailing list.
It is entirely different between listening explicitly at the 'telnet'
port which is only used for logins and capturing the first 200 bytes of
traffic each way, and monitoring the 'smtp' port and having to capture
ALL mail that is transferred to the site. A big difference in content,
e.g. having to scan thousands or tens of thousands of messages for a
credit card number.
> } I do believe being able to have settlement over the internet so that
> } money transfer issues can be resolved is a necessary issue if we are to
> } use the Internet for more than just data.
>
> We agree 100% --- my very point.
>
>
> } However, for *some* people who use less, they *may* find that metering is
> } cheaper than flat rate. But I suspect that metering is used as a means
> } to make those who use more pay more.
>
> Indeed... I think so too. But that strikes me as just right: why
> *shouldn't* people who use more pay more?
Because the changes are never made to adequately apportion costs; they
have, every time they were done, used to soak everyone for much higher rates.
> } There are some uses that require huge bandwidth for short periods.
> } Metering by usage rather than by pipe size means that one can't try to
> } demonstrate these things; if it costs 10 times as much for use of a 1.4
> } mb T1 as opposed to a 56K line (when common pricing sets a T1 at about
> } 150% of a 56K line) then it becomes very expensive to try tests. And I
> } suspect this is exactly the sort of thing that will happen.
>
> I don't understand your point. almost no one wants or needs 'fixed
> bandwidth' service --- just about EVERYONE can get by quite well
> with sporadic service --- they could be being _charged_less_ because
> *someone*else* could be using their 'reserved capacity' while they're
> not. [How long each day can you SPEND sitting in front of Mosaic
> having GIF files pumped at you?]
Ah, packet switching, X.25. Every place it's used, it's used sparingly and
with care because it's *very* expensive.
> } : Only because you have barely a clue about how the Internet works
> } : today. You seem to REALLY think that the 'hidden' network
> } : bandwidth is unbounded and free, and so it should make no
> } : difference if you
> } : a) don't use the internet at all
> } : b) telnet someplace [and so eat up a few kbytes per
> } : minute of bandwidth]
> } : or c) X-windowing to the same place [and so eat up a hundred
> } : kilbits per *second* of bandwidth].
> } : Same cost for all three situations from the flat-rate fans.
> }
> } Not true. If I do X-Windowing on a regular basis, I'm going to find a
> } 14.4K slip connection is too small, and probably move up to a 56K.
>
> We've tried this and it is tolerable but not great. X-windows was
> designed assuming that the server and client were sharing a high-bandwidth
> LAN and so is *incredibly* profligate with network bandwidth.
But that's another point. The user doesn't know or may not have control
over how much bandwidth the applications he uses take. If he pays by
usage, a resource hog application could end up sucking out a month's
budget in a short period of time.
> } : What costs is having *available*resources*, and sooner or later [I hope
> } : sooner] all the various players will start properly accounting-for
> } : and billing-for them. Right now, for most parts of the Internet the
> } : infrequent/efficient/sparing users subsidize the gluttons. I look
> } : forward to the day when everyone pays their fair share, so that the
> } : small users get truly tiny bills, while the bandwidth-gobblers whose
> } : rapacious appetites have contributed so much to the need for all the
> } : T3 and T1 capacities will start PAYING trhough the nose.
> }
> } When Illinois Bell used the same rationale to eliminate larger free
> } calling areas, what we saw was that the sparing users of phone service
> } saw very tiny to insignificant drops in their phone bills (like a phone
> } bill going from $15.00 a month to $14.50), while heavy users saw massive
> } 300% to 500% increases.
>
> Using the highly-regulated local telco monopolies as examples of
> proper-economics is not very good. I can call Seattle [3500 miles away]
> for less [by half!] than I can call Blacksburg, 40 miles up the road.
> All that tells me is that the local monopoly that C&P has on my
> phone line is screwing me, but it tells me _nothing_ about the underlying
> economics.
Exactly. The point being that the system we have of non-settlements
providers does seem to work and do so in a manner that brings us
not-very expensive data pipe. Start charging by volume and the costs
for usage will skyrocket.
> } : Usenet is probably an example of something I hope will change *SOON*.
> } : My provider carries something like 5500 newsgroups and moves well
> } : over 100megabytes a day. *I*PAY* for my provider having to have
> } : a fat enough pipe [and, of course, in turn HIS provider having a fat
> } : enough pipe and all the way up to the backbone having fat enough pipes]
> } : to move all of that mostly-crap around. It'll be a _great_ day
> } : when there is some way to participate in the Internet and _not_ have
> } : to pay implicitly for all of that bandwidth that is gobbled by
> } : that stuff that is of almost no interest to most folk.
> }
> } But you can get newsgroups by satelite for $1800 installation and $30 a
> } month after the first year. Using this route, all that a site needs the
> } *huge amounts of bandwidth* are for live transmissions and FTPs.
>
> All true. So what's the point? And what about MOSAIC, weather-GIFs
> in real time, FTP, etc? And usenet, itself, doesn't use all that
All it takes to run the 5000+ newsgroups on Usenet is *one* 56K pipe.
It's because people want fast transmissions of other activity (such as
FTP and Telnet) that they need this kind of capability.
> } Do you *really* think that an increase in the cost of taking a full feed
> } of newsgroups is going to make your internet service provider cost less?
> } Guess again. You might see a 50c a month drop, but everything would be
> } meter, meter, meter and you couldn't afford to do much more. In the end,
> } it would be much more expensive.
>
> Methinks you know little about how the systems work --- you make it
> sound like it is an electric meter and some guy is going to have to
> drive to my house to look at some register in my computer.
No, the provider will look at the meter at *his* computer.
> The
> metering is largely *done*anyway*, and the actual traffic source is
> trivially pinpointed [you think it would be hard or expensive for
> the daemon to know which process handed it an IP packet?]. The
> actual billing and such is already done, and so the additional
> processing is pretty trivial [as I say, I've already writted the
> code to add this to an accounting/billing system once]
Do you know how many packets travel over the Internet? I think someone
said that just one segment of the national backbone had traffic
approaching 2 trillion packets, 50 billion bytes, in a single month.
This does not include the probably heavier local traffic that never
touched it or that used other networks.
If the routers have to count and settle (assign charges for) those
packets the router is going to be slower which means it needs to be a
more expensive, faster router.
>
> As for the cost/provider, yes I think it would cost less to have a
> system that had [say] slow FTP's because it was hanging on just a
> 56K line instead of dual-T1s. I think it would cost less to have a
> system that _didn't_permit_ the high-bandwidth real-time
> applications [even, perhaps, including NOT supporting FTP or
> supporting it only in some background/slow fashion [e.g., FSP]].
> Why do you think it wouldn't? Why wouldn't setting up a service
> that could support X customers with just a 56Kb link not cost less
> than one that supported the same number of customers [but with bigger
> appetites!] on a T1 link?
Certainly a 56K link costs less than a T1. And it might cost less. We
have the situation here. Digital Express runs a T1 (I immediately
noticed the difference when they had 56K service and the huge improvement
in speed). Clarknet down the road in Columbia has a 56K, and the
difference is about $5 a month.
>
> And you shoudl be a bit more circumspect about throwing out numbers
> [like "50c a month"] into an arena where you couldn't possible know
> anything about what anyone is paying.
These are the numbers that occurred when Illinois Bell changed the local
calling plans so that flat rate service was eliminated. Massive
increases for some, significant increases for the rest. (One reader from
Chicago indicated that he could find _no one_ in his circle of friends
who had seen his rates go down, even though this is what Illinois Bell
claimed would happen for MOST USERS. That year saw RECORD PROFITS for
Illinois Bell. I see nothing wrong with a company making money, but not
by intentional misrepresentation.)
Reports on Security Problems: To Subscribe write PROBLEMS-REQUEST@TDR.COM
Paul Robinson - paul@tdr.com / tdarcos@MCIMail.com / tdarcos@access.digex.net
Voted "Largest Polluter of the (IETF) list" by Randy Bush <randy@psg.com>
Voted "Largest Polluter of digex.general" by Mike <voss@orange.digex.net>
