t byfield on 24 Feb 2001 19:36:17 -0000


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

<nettime> [cook@cookreport.com: Light, IP and Gigabit Ethernet]


ex-nettimer gordon cook suggested that this might be of interest 
to some people on the list, and indeed it should be: back when
dogs r00led the earth, there was some 'mapping the power' noise,
an impossibly complex subject then, and even moreso now. sadly,
what the abstract below doesn't talk about in any detail is the
emphasis on 'condominium' models of collective/community owner-
ship of fiberoptic networks in canada's 'canarie' project--quite
distinct from the 'market-based' obsessions of the US, which so
far have served primarily to perpetuate the hegemony of the big
telcos (the 'baby bells' and their competitors arising from some
'liberalizations' of the 70s and 80s--sprint, MCI/Worldcom, etc.).
there's a strong argument to be made that much-ballyhooed 'P2P'
networks will only succeed in a meaningful sense if the infra-
structure's ownership, hence maybe management, is likewise P2P--
community-owned.

cheers,
t

----- Forwarded

Date: Mon, 19 Feb 2001 21:44:28 -0500
To: tbyfield@panix.com
From: Gordon Cook <cook@cookreport.com>
Subject: Light, IP and Gigabit Ethernet A Road Map for Evaluation of
 Technology Choices Driving the Future Evolution of Telecommunications


Contrary to some opinions, the COOK Report finds that the Internet 
revolution is not spelled dot com.  The revolution is in fact to be 
found in a total revamping of the transport of bits.  While the dot 
com empires of 1999 collapsed in 2000 the cost effectiveness of 
pushing the Internet Protocol over glass yielded more dividends than 
ever before.

A growing amount of telecom traffic has migrated to a growing amount 
of fiber.  The pure Internet play throws out SONET effectively 
doubling available fiber in the case where  redundant loops were 
used.  Whereas lighting each new fiber used to call for new bays of 
OC-48 SONET equipment at $100,000 a bay and up, a strand can now be 
lit at a gigabit by a $7,000 Ethernet switch on each end. 

While gigabit Ethernet over glass is the current preferred Internet 
way, ten gigabit Ethernet transport will be arriving by year's end. 
If 40 lambdas per strand were high end in 2000, 160 is likely to be 
common by year's end.  With the completion of multiple metro fiber 
build outs, end-to-end fiber may now be taken for granted by most 
business customers.  The explosion of bandwidth as the result of more 
fiber and technology that squeezes more bandwidth from each strand 
has meant that, in some instances, the delivery of a gigabit costs 
about what a T-1 did a decade ago.

The bottom line is that telecommunications which is prepared to 
forego traversing the legacy PSTN is now upwards of 1000 times 
cheaper than that which powers a circuit switched voice call.  While 
corporate managed VPNs have been able to avoid the PSTN for some 
time, a new development has emerged in Canada where customer 
management of optical wavelengths using the OBGP protocol holds the 
promise that by year's end users of Canada's new public sector 
gigabit Ethernet over fiber infrastructure will be avoiding carrier 
clouds entirely.

At the basic levels of both transport and network management the 
Internet revolution is shaping up to tell the PSTN that it is no 
longer needed.  In telephony meanwhile protocols are being developed 
that will allow the diversion of large amounts of PSTN traffic to the 
Internet.  ENUM is the major such protocol.  This will allow Internet 
carriers to offer and deliver many services to PSTN attached phones 
that the PSTN itself cannot negotiate.  Other protocols such as 
instant messaging are shaping up as coordinators for PSTN activity 
and off on switches that can control Internet connected devices.

Fiber to the home is becoming more common and companies like World 
Wide Packets are gearing up to make gigabit Ethernet termination 
equipment that will give connected families, telephone, fax, high end 
video, ordinary TV and data off of the same line.  Canarie the 
Canadian advanced internet agency has some interesting ideas about 
these developments stating that Divergence rather than Convergence 
may be the key to low cost fiber to the home.  Here is a narrative 
paraphrase of the language of a slide from the presentation 'Optical 
Communities' in September 2000.

When people first started looking at Fiber to the Home (FTTH), they 
deemed it to be too expensive because it assumed all services would 
be converged - date, voice and video. They noted that expensive 
terminal equipment would be required to segregate voice, data and 
video services at the home. Meanwhile voice traffic has largely gone 
wireless. Note that lifeline voice can significantly increase system 
costs by demanding high reliability and depending for this on DC 
battery power, 911 services. Perhaps it is time to conclude that the 
big driver for residential broadband is not voice or video.  It is 
the Internet. Very soon Internet will carry video and second line 
voice.  So instead of building a converged network such as FSAN, HFC, 
etc build an Internet network only.  Divergence rather than 
Convergence may be the key to low cost FTTH.

While the power of the new systems is awesome, there are additional 
issues that will keep very interesting the life of anyone who must 
evaluate these changes and plan a winning strategy for the future. 
While one better be aware of the key differences in the power of the 
technology when compared to the circuit switched world's way of doing 
things, one also needs to understand that progress has, in this case, 
waded out into new and uncertain territory. There are some growth and 
scaling issues where the answers are not yet clearly understood.

For example readers should consider Bill St. Arnaud's paper on 
scaling issues of Internet growth. < 
http://www.canet3.net/library/papers/scaling.html>  If the 
suppositions in this paper prove to be correct, then the role of 
backbones will have to be rethought and much Internet topology 
rearranged. One of the developments that influenced his thinking 
occurred on December 4, 2000 when Essex Corp announced an advance 
that enabled it to push as many as 4,000 lambdas down a single strand 
of fiber.  On December 8 St. Arnaud posted the following to his 
Canarie mail list.

"[Here are some excerpts from a recent article in Lightreading that I 
believe illustrates the point that Ultra dense wave division 
multiplexing systems is not about bandwidth, but connectivity.  A 
number of companies are starting to realize that today's Internet 
architecture (which is still fundamentally based on the old telco 
network model) will not scale. As intelligence moves to the edge, the 
existing network architecture must grow at some function related to 
the square of the number of users.  However [the problem is what to 
do] if the increase in the number of the customer's own wavelengths 
and the fiber in the network only grows linearly? Intriguingly such a 
customer owned network architecture starts to closely resemble the 
neuron architecture of the brain.  Perhaps mother nature long ago 
discovered the most efficient architecture for distributed 
intelligence?  - BSA]"

 From the December 4, 2000 issue of Lightreading.... 
http://www.lightreading.com/document.asp?doc_id=2760

"We don't need thousands of wavelengths for bandwidth, we need it for 
connectivity," says Simon Cao, chief  technology officer at Avanex 
Corp. (Nasdaq: AVNX), a company that's  developing high channel-count 
wavelength systems. Cao figures that the availability of thousands of 
channels in combination with tunable lasers will make it possible to 
eliminate complex optical switches, by using wavelength routing 
instead . That idea is the driving force behind all of Avanex's 
developments, he says."

"Despite its careless capacity talk, Essex has also thought out how 
to take advantage of the extra connectivity. Moodispaw reckons that 
Essex's solution will be perfect for customer-owned wavelengths in 
the access network. About 1 Gbit/s would be adequate to supply a 
gigabit Ethernet line to a business. Each customer pays less for its 
wavelength, but the service provider is able to sell to a lot more 
customers, so everybody is happy."

What this means is that we likely soon see customer owned wavelengths 
of a gigabit. Such wavelengths will be plentiful, affordable and 
matched to the speeds of the Internet's basic gigabit Ethernet 
end-to-end architecture. When only the rich and mighty carriers owned 
fiber and the very expensive lasers that could pump bits and the even 
more expensive SONET equipment that sustained the  bits, we imagined 
that bandwidth was a valuable commodity obtainable only from "on 
high," or from upstream. Indeed the carriers recognized their 
position and charged large sums for the scarce commodity that they 
delivered.

This situation has dramatically changed.  If you own a piece of fiber 
in a metro area network, you can power that fiber with a lambda 
delivered as gigabit Ethernet for a one time investment of about 
$15,000. When you run out of bandwidth, adding the multiplexer 
necessary to turn the one lambda into four is also affordable.  In 
this sense bandwidth can now be generated by customers and used very 
cheaply at the edge of the network over hops that in ideal conditions 
can be as much as 100 kilometers.  Suddenly these customers can throw 
cheap bandwidth at quality of service issues.  Their whole outlook on 
life starts to change rapidly.  George Gilder was observed to say 
that "the companies that exploit bandwidth recklessly, will win."

On January 29, 2001 Bill St. Arnaud observed "Although I may disagree 
with Gilder on some of his predictions, I think he's gotten this one 
right.  To date most advanced optical network research is based 
around the assumption that bandwidth is a precious and rare commodity 
and therefore we must optimize the network topology and try to 
extract every possible bit of capacity.  One of the underlying 
assumptions of the existing CA*net 3 network and proposed CA*net 4 
network architecture is that bandwidth can be wasted.  Rather than 
concentrating on applications and technologies that optimize the use 
of bandwidth, we want to concentrate on applications and technologies 
that waste bandwidth but in doing so derive the maximum benefit for 
the user - hence our focus on "customer empowered" optical networks. 
There is no question in our mind with customer empowered networks 
that bandwidth will be wasted, wavelengths will be orphaned and seen 
from the perspective of a central carrier the network topology will 
be less than optimum.  But in a world of thousands of wavelengths and 
near infinite bandwidth who cares?  The true measure is whether the 
customer can easily and rapidly deploy new broadband applications and 
services."

"At CANARIE we are initiating a number of projects with this premise 
in mind - OBGP which will allow customers to setup and tear down 
their own wavelengths at will, WDD- wavelength disk drives will drive 
new concepts in distributed super computing, object oriented 
networking enables the wavelength to become a software object, 
community based condominium fiber networks, and much, much more. Stay 
tuned for more announcements and developments in this area."

Arnaud's words portend a seismic shift in the telecommunications 
landscape. With the exception of  some important qualifiers we 
believe that he is correct.  One must understand that, for the time 
being, bandwidth is cheap only in proportion to the distance that it 
must travel. Very high bandwidth sent over long distances is still 
quite expensive. Bandwidth that must be provided on a corporate wide 
scale for mission critical activities is also somewhat expensive. 
What Arnaud is talking about will change the world.  The only 
question is how fast.

To sum up.  Costs are falling.  Depending on the regulatory 
environment, the amount of capital necessary to become a small scale 
telecommunications provider is rapidly shrinking.  Independent back 
yard gigabit Ethernet providers can deliver broadband services at a 
fraction of the cost but equal in quality to what is accessible from 
larger more traditional companies.

There appears to be a choice of two paths to our telecom future. One 
is to go with the highly innovative pure Internet approach of gigabit 
Ethernet over condominium fiber. Such a choice empowers the customer, 
facilitates decentralization over centralized control and provides 
small and innovative businesses with the environment that they need 
in order to flourish.  The other path is to try to fore stall the 
innovation by squashing competitors with a massive vertically 
integrated company founded on older technology and leveraging access 
to content and over a network monopoly so pervasive that people will 
find they have no choice but to buy it.  What could be in store for 
us all, if things go in this direction, was summarized by Scott 
Cleland CEO of the Precursor Group on Friday January 19th, 2001. 
"Precursor believes AOL-TW has budding 'Microsoft-like' potential to 
grow increasingly dominant by being the leading national company that 
brings together the various online interfaces (content, Internet 
access, buddy lists, instant messaging, etc.) to become the de facto 
consumer online access market standard much like Microsoft Windows 
brought together the various desktop applications to become the de 
facto consumer software market standard."

On the one hand, the AOL infrastructure, which, compared to what the 
Canadians are doing, can most kindly be called archaic, has to be a 
terribly attractive vision for the ILECs.  After all, it will give 
them plenty of time to finish depreciation of their copper plant.  On 
the other hand innovative Internet technology ventures have the most 
to gain from taking the Canadian road.  To facilitate making this 
choice with more certainty, in this publication we offer a 
recapitulation of COOK Report interviews on the shifting market 
domain of the optical network.  These interviews are unique in that 
they treat complex key technical issues with attention to depth and 
detail found no where else.  We believe that they form a set of 
resources that will permit readers to decide for themselves what 
direction is most beneficial to the future of their enterprise.

As we ponder these developments, we realize that we may be standing 
on the cusp of events that may overturn the economic structure of the 
first five years of the commercial Internet.  The largest backbones, 
UUNET and Sprint, set the price of bandwidth and determined 
conditions for connection to the Internet.  Last week WorldCom 
announced that it was debranding mighty UUNET and Mike O'Dell, the 
chief architect of the Internet's largest global backbone, left the 
company.  The problem is that these services take more and more 
resources to pay for ever larger routers and support huge growth 
rates in traffic.  It is beginning to look like revenue derived from 
running such a facility may no longer match expenses as the 
availability of alternative sources of cheap bandwidth and Internet 
transit increases.  For the largest backbones sources of power only a 
year or two ago may now be turning into liabilities. The economics of 
bandwidth are in flux to such an extent that there is growing 
suspicion that for many large players business models that were 
sustainable five years ago have lead to the acquisition of huge debt 
today and that the next year may bring a round of major bankruptcies 
as the new economies of the technologies described in this report 
render traditional business models based on old technologies 
unsustainable.

In the end the path of the internet as the stupid network driven by 
these new technologies will predominate. We begin this six part 
report http://cookreport/lightipgige.shtml with a section on the 
basic Internet architecture and technologies of "Light, IP, and Gig 
E" to be followed by a section on their adoption in Sweden, Holland 
and of course our long special issue on Canada.

The report may be ordered by clicking on 
https://secure.fast.net/cookreport/order.html.

-- 


****************************************************************
The COOK Report on Internet, 431 Greenway Ave, Ewing, NJ 08618 USA
(609) 882-2572 (phone & fax) cook@cookreport.com   Index to 9 years
  of the COOK  Report at http://cookreport.com          For info on new report
go to  http://cookreport.com/lightipgige.shtml .  'Light IP & Gig E"  serves
as tutorial on on going economic model of Internet infrastructure - $375.00
****************************************************************

----- Backwarded

#  distributed via <nettime>: no commercial use without permission
#  <nettime> is a moderated mailing list for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: majordomo@bbs.thing.net and "info nettime-l" in the msg body
#  archive: http://www.nettime.org contact: nettime@bbs.thing.net