The Great Exchange

Michael Shields writes:

In article <199805271744.NAA21312@jekyll.piermont.com>,
> In the long run, why are we assuming there will be locality of
> traffic?
>
> It is true that the old PSTN has locality of traffic, but it doesn't
> have flat rate pricing, or the usage patterns that the Internet has.

Why are you assuming that the Internet will continue to have
non-distance-sensitive pricing, when it clearly has distance-
sensitive costs (ultimately)?

I'm assuming it for several reasons.

1) The price of metering packets by destination exceeds the cost of
providing the underlying service as it stands. This does not make me
think that distance sensitive pricing has a bright future.

2) Long distance carriers have essentially eliminated distance
sensitive pricing for domestic U.S. calls, and at the moment the price
on such calls is only a hair above the local carrier settlement costs
of 7 cents per minute (3.5 cents per side.) Cellular carriers are
getting into the act, eliminating long distance charges for a small
flat fee. International pricing is also plummeting. None of this leads
me to believe that the future for metered PHONE SERVICE is very
bright, and you're trying to argue that metered internet service has a
bright future?

3) It does not appear that moves towards metered pricing have been
well recieved by consumers. ISPs are sticking to their current pricing
model, although they are often limiting the flat rate zone to
"reasonable" personal usage levels of a couple hundred hours per
month.

In short, I see no economic basis for such metering, and no moves
towards such metering. Maybe I'm wrong, but I doubt that we're going
to see any change in this any time soon.

Perry

Michael Shields writes:

[...]

> Why are you assuming that the Internet will continue to have
> non-distance-sensitive pricing, when it clearly has distance-
> sensitive costs (ultimately)?

I'm assuming it for several reasons.

1) The price of metering packets by destination exceeds the cost of
providing the underlying service as it stands. This does not make me
think that distance sensitive pricing has a bright future.

I don't think I can agree that the cost of metering is unreasonable,
assuming you have the ability to monitor the traffic on your lines to
know what sites are using the bandwidth. This data is available today
on routers capable of flow-export; I think it will need to be a
standard feature on all major routers in the future, because it is
very valuable data even if not used for implementing a distance-
sensitive pricing model.

What you would need to determine is your cost per GB to each
destination prefix (or from each source prefix) outside your network.
You can do that by determining the cost of each of your backbone and
border links, and the makeup of traffic on them. This computation
scales with the number of links in your network times the number of
prefixes in the Internet.

I don't think the cost of calculating this would exceed the cost of
providing the underlying service. It may or may not be "cheap enough".

Please let me know if I'm misunderstanding something.

2) Long distance carriers have essentially eliminated distance
sensitive pricing for domestic U.S. calls, and at the moment the price
on such calls is only a hair above the local carrier settlement costs
of 7 cents per minute (3.5 cents per side.) Cellular carriers are
getting into the act, eliminating long distance charges for a small
flat fee. International pricing is also plummeting. None of this leads
me to believe that the future for metered PHONE SERVICE is very
bright, and you're trying to argue that metered internet service has a
bright future?

It's one thing to have semi-unmetered service, of which there are
plenty of examples in phone networks -- unlimited local calling, or
unlimited domestic calling. It's very much another to have fully
unmetered service: free calls of any length to anywhere in the world.
Until I can make a one-hour call from DC to Moscow for the same price,
or even a similar price, as a one-hour call to New York, it's not
equitable that I can ftp for the same price.

3) It does not appear that moves towards metered pricing have been
well recieved by consumers. ISPs are sticking to their current pricing
model, although they are often limiting the flat rate zone to
"reasonable" personal usage levels of a couple hundred hours per
month.

In short, I see no economic basis for such metering, and no moves
towards such metering. Maybe I'm wrong, but I doubt that we're going
to see any change in this any time soon.

Could be. I think the economic basis is there; I just don't know if
it's strong enough to overcome customers' love of flat-rate.

As long as usage is mostly homogenous within a price class, there
won't be any incentive for non-flat-rate pricing. If a substantial
number of users start doing daily large international data exchange,
that may start to hurt. The question is whether a relatively small
percentage of users account for a significant fraction of the traffic
on the expensive lines. I don't know the answer but I would not be
surprised to find many customers have much more long-haul traffic than
average.

Probably it will never make market sense to have distance-sensitive
traffic pricing for "low-speed" users, where the cost of providing the
service is mostly the cost of tech support, billing, dialin or xDSL
aggregation, &c., and bandwidth is a small proportion of the cost.

Until I can make a one-hour call from DC to Moscow for the same price,
or even a similar price, as a one-hour call to New York, it's not
equitable that I can ftp for the same price.

Which resource is it that you are pre-empting, Michael, when you FTP
that file?

Circuit switching (TASI notwithstanding) and packet switching have
different scarcity, and therefore pricing, models.

Could be. I think the economic basis is there; I just don't know if
it's strong enough to overcome customers' love of flat-rate.

I think that it is not, and I don't think it's worth it to a whole
bunch of people to buy three times as much router CPU to do the work.

Probably it will never make market sense to have distance-sensitive
traffic pricing for "low-speed" users, where the cost of providing the
service is mostly the cost of tech support, billing, dialin or xDSL
aggregation, &c., and bandwidth is a small proportion of the cost.

With current trends, will the actual bandwidth _ever_ be more than a
small fraction of the cost? Except _maybe_ on trans-oceanic lines...
and bet on that situation to get better, not worse, as well.

Cheers,
-- jra

In article <19980528101417.10618@scfn.thpl.lib.fl.us>,
[...]

> Probably it will never make market sense to have distance-sensitive
> traffic pricing for "low-speed" users, where the cost of providing the
> service is mostly the cost of tech support, billing, dialin or xDSL
> aggregation, &c., and bandwidth is a small proportion of the cost.

With current trends, will the actual bandwidth _ever_ be more than a
small fraction of the cost?

Despite preductions, very few resources have ever actually become "too
cheap to meter".

Can you back up your comment about metering requiring three times the
current CPU capacity of routers?

No, I pulled it outta my butt. But I do hear that a significant amount
of router CPU is necessary to add this function, particularly at high
packet rates, and a similar amount of analysis CPU would then also be
necessary.

Cheers,
-- jra