William H. Magill on Tue, 2 May 2000 13:27:07 -0400 (EDT)


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

[PLUG] DSL costs and pricing issues


>   > > "$15/month Netaxs charge, $40/month Bell charge for 640k down, 90k up.
>   > >  $40/month Netaxs charge, $60/month Bell charge for 1.6M down, 90k up. "
>   > > 
>   > > Why do they charge more, just for a faster connection? I can maybe
>   > > understand Bell charging more for higher bandwidth, but why the ISP?
>   > 
>   > Because you'll have the ability to use more of our available bandwidth,
>   > which isn't infinite, and costs us money.  It's the same reason why we
>   > charge more for a T1 than we do a T3, and why collocation charges are
>   > based on bandwidth utilization.
>
>   You charge *more* for a T1 than a T3 because of *bandwidth utilization*?

Communications pricing has rarely had anything to do with actual costs. 
Pricing always been viewed a way to regulate demand.

Today's pricing games are all artifacts of the fact the telecom people have
no idea how to live in a competitive market. They still think they are in a
regulated market.

T1s were (and still are) priced high because they were priced back in the
days when everything was regulated. There was a well defined ritual
involved in pricing in front of the PUC. The RBOCs were encouraged to find
as many costs as possible to associate with a product when they made their
tariff requests because they would likely only get a portion of their
request. That made the regulators look good -- they were cutting costs; and
the made RBOCs look "subservient" to the public interests. 

The end result was that the total  costs associated with a product was
divided by some number of probable purchasers at some ridiculously high
price, and it therefore resulted is a large number being divided by a small
number which resulted in the fact that a T1 line costs 10 or 100 times 
times what the real cost to provide the service is.

Now you say - wait a minute, that can't be true. But look at the facts.
A T1 line costs around $400 per month to one CO (there are distance
issues.) A DSL line costs $49 per month to one CO. What is the difference
between the two? The T1 line is 1.5 meg baud, the DSL is is 1.6meg baud.
Oops...Besides cost, what is the difference in the service? Hmmm...

The next term is "Dis-intermediation." Bell has LOTS of T1 customers paying
$400 per month. Suppose they allowed that same 1.6 meg up-link speed. All
of s sudden, bell looses its T1 customers to DSL. A WHOPPING BIG loss in
revenue. Bell is scared shitless about loosing that revenue. They never
heard of the "big box" theory of retailing.

No ISP out there today ANYWHERE IN THE WORLD is capable of providing the
kind of bandwidth needed to support the connections they are currently
selling. That is the reason that no ISP will offer you any kind of
connection guarantee [AKA, Service Level Agreement] that is not riddled
with "circumstances beyond our control" clauses. (And by ISP, I am
including the "big boys" MCI, UUnet, Sprint, Worldnet, Cable and Wireless,
etc. Local ISPs are just a change in scale.) They all know that except for
very tightly controlled situations, the existing demand for bandwidth far
exceeds current and projected capacities. Cost has nothing to do with it.
The capacity simply isn't there at any price -- today. All ISPs, and RBOCs 
work on something called a "fill rate" -- their "best guess" (or
prediction) as to how many of their "subscribers" will really be using the
service. They work from the bottom up - buy a pipe, fill it up, buy a
bigger pipe. 

This is very different from the Electric Utility business, which bases its
pricing strategy on "peak load demand." And whose plant is based on the
worst case scenario -- what happens when everybody turns on the TV at the
same time. (Like after a power outage.) They MUST be capable of providing
for that instantaneous peak or they would never get back on line.

Why on earth should co-location charges be based on bandwidth?
Co-location means square footage necessary to physically locate equipment.
Then you string some circuits between your location and somebody else's
location, nominally an RBOC. Co-location actually DRAMATICALLY cuts the
cost of those circuits because you are no longer dealing with long wires,
but just a bunch of short jumpers. Yes, it is true that in order for the
RBOC to provide you with a circuit that they have to install equipment
that they would not otherwise install. But that is a capital cost of doing
business. That cost is (supposed to be) borne by the stock holders. It is
the investment made to enable them to do business.

In communications, Pricing is done to control demand and to a lesser extent
control usage. It is only very recently that the concept that "lower price
means more sales" has begun to creep into the industry. Cellular telephone
rates are a good example. There were two problems associated with the slow
initial consumption of "Cellular minutes." First they were expensive, and
second, they were priced based on "actual usage." Pricing based on actual
usage sounds good at first. But it means that your bill is very, very
volatile, and with very few exceptions, it means that you really have no
control over that volatility -- especially when the charge per-unit of
usage is high. So ATT pioneered the "one-rate" scheme in Texas -- usage
and "market penetration" went through the roof!

Global Crossing did the same thing with "dark fibre." They strung Fibre
Optic cables across the Atlantic and sold them for a small fraction of the
rates being charged by everyone else - they have been incredibly
successful. While others have unsold capacity, Global Crossing is laying
more as fast as it can -- and that all based on the old Fibre technologies
of 2 years ago!

Today, finally, there are options in the T-circuit game. PECO Hyperion has
its own cable and fibre plant as do others, and the result is that
T-circuit costs are finally dropping.

So If I want a DSL circuit from my house to someplace else, I only have to rely
on Bell for that last mile (today, but maybe not forever.) If I pick COVAD
as my DSL provider, they buy a "dry-copper" pair from Bell from my home
back to my CO. In that CO, Bell connects that circuit to COVAD's "switch"
in COVAD's "co-located" corner of the building. There is no bandwidth
involved in the transaction. Strictly hard-copper from my home to COVAD.
Next COVAD links their CO equipment to the MCI POP via a PECO Hyperion
Fibre Circuit. That PECO circuit is NOT terminated in Bell's part of the
CO, but rather in COVAD's.... Bell has nothing to do with the cost, except
to charge some rent for the conduit. (What they may try to get away with
charging for is a separate issue.)

I've been somewhat simplistic (but not much), and have glossed over some of
the "risk factors" normally included. But the pricing game is changing fast. 

George Gilder describes the new era -- "infinite and free" -- Virtually
Infinite Bandwidth at virtually no cost. It's not here yet, but it will be
in 3-5 years. Companies are having a very difficult time with the reality
of the pent-up demands for bandwidth. The Bell Atlantic DSL rollout has
been a glorious case study of how to underestimate demand -- you might
think they were taking lessons from Apple. Price is much, much less an
issue than performance, and it has been now for at least 5 years.

Talk in the industry is that Bell Atlantic will in fact, up their default
up-link speed from 90K to 1.6 within the next 3 months as they upgrade to
the next generation of DSLAMs.

The latest issue of CE-Pro, a magazine for the Custom Electronics
installation industry, has an article about why any CE shop should buy a
DSLAM from Cisco (the 6400) for $13K and go into the DSL business -- probably
supports 64 or 128 ports in the base model, expands up to 14,000! Quite
an interesting box with fascinating capabilities, reportedly at 10% of
NORTEL's price!
        http://www.cisco.com/univercd/cc/td/doc/pcat/6400.htm

-- 
                        www.tru64unix.compaq.com
                              www.tru64.org
                             comp.unix.tru64
                        
T.T.F.N.
William H. Magill                          Senior Systems Administrator
Information Services and Computing (ISC)   University of Pennsylvania
Internet: magill@isc.upenn.edu             magill@acm.org
http://www.isc-net.upenn.edu/~magill/

______________________________________________________________________
Philadelphia Linux Users Group       -       http://plug.nothinbut.net
Announcements - http://lists.nothinbut.net/mail/listinfo/plug-announce
General Discussion   -   http://lists.nothinbut.net/mail/listinfo/plug