Transcript presentation - NORDUnet Networking Conferences

Lightwaves at the end of the telecom
tunnel?
Nordunet Annual Conference
Reykjavik, August 24th 2003
Yves Poppe
Dir. IP Strategy
1
Agenda

Evolution of Transoceanic Internet Capacity Demand
and Supply

Aftermath of the Great Telecom Storm

The R&E World Sees More and More Light

Next (Light)wave of Opportunities
2
The pitfalls of predictions and forecasts
3
In October 1994 Teleglobe and its partners
inaugurated Cantat-3 with two fiber pairs,
capacity of 5 gigabit (2x2.5Gb) linking
Canada to the UK, Germany, Denmark,
Iceland and the Faroe Islands.
Doubled the capacity under the Atlantic
155mb was earmarked for data
Engineering estimated 17years to fill the cable
4
183
45 Mb/s (in trial)
155 Mb/s CT-3
Iceland PTT
ATM Netw ork
KDD/NTT ATM R&D
Netw ork
(JAPAN)
34Mb/s
Ibaraki, Japan
Lake Cow ichan
Pennant Point
Vancouver.BC
45Mb/s
B-WIN
(DFN)
45Mb/s
155 Mb/s CT-3
Teleglobe
ATM Test
Network
34 Mb/s
Montreal
DT
ATM Netw ork
45Mb/s
34Mb/s
CANARIE
CA-Net-2
Netw ork
(CANADA)
45Mb/s
STARTAP
(Chicago)
Berkom
SWITCH
(Sw itzerland)
JAMES
34Mb/s
45Mb/s
Sprint
ATM Netw ork
(US)
SuperJANET
ATM Netw ork
(UK)
34Mb/s
34Mb/s
RENATER
ATM Netw ork
(FRANCE)
SIRIUS
ATM Netw ork
(ITALY)
45Mb/s
MCI vBNS
Netw ork
Cantat-3 and R&E 1995-1998
5
How Reality Turned Out to Look Like






The internet tsunami took everybody by surprise.
Cantat-3 was full in less than 3 years.
The magic potion of DWDM : five years later cables of 1000 times the
capacity of Cantat-3 were being installed.
Deregulation, easy access to capital, advances in laser and fiber
technology and spectacular internet growth created a new generation
of global cable builders: Global Crossing, Level3, FLAG ,
360networks and resulted in a cornucopia of transmission capacity.
R&E transatlantic connectivity : from kb/sec to meg/s to gig/sec in
less than 10 years
After 3-4 years of spectacular growth, a peak in early 2000 and a
long steep downhill in the telecom industry.
6
The Battle of the Atlantic

Design capacity and RFS
–
–
–
–
–
–
–
–
Gbps* RFS
Level 3/GC (Project Yellow)
TAT-14 (Club)
Hibernia (360networks, Inc.)
FLAG Atlantic-1 (FLAG/GTS)
Atlantic Crossing -2 (Global Crossing)
TyCo Global Network
Apollo (C&W)
Total
Lit capacity early 2003: 2,338Gb
(source: Telegeography)
1,280
Sep00
640
Apr01
1,920
Jun01
2,560
Sep01
2,560**
1Q01
2,560
Jun02
3,200
Feb03
12,160Gbps!
* = Design capacity
** = Cancelled, AC-2 joining Level 3
7
The Battle of the Pacific

Design capacity and RFS
–
–
–
–
–
–
–
–
–
Gbps* RFS
TPC-5 (club)
20
Dec98
Southern Cross
480
Nov00
China-US (club)
80
Jan 01
PC-1 (Global Crossing & Marubeni) 640
Apr01
Japan-US (club)
640
Oct01
Tyco Pacific
5,120**
Jan03
FP-1 FLAG Pacific
5,120*** 2Q02
360 Pacific
4,800*** 3Q02
Total 6,980Gbps
* = Design capacity
** = april 01: Tycom joins FLAG
aug 01: FLAG withdraws, Tycom continues alone; RFS postponed
***= project dropped
Lit capacity early
2003: 1,043
(Telegeography)
8
Transoceanic buildout frenzy completed

With the activation of the C&W Apollo transatlantic
and Tyco’s transpacific cable the current phase of
intense build-out is coming to an end

With current fill rates low and about 3 years between
start and completion of a project, this means new
cables unlikely before 2007-2008

Weak point remains Europe-Asia capacity. Should
improve with SEAMEWE4 scheduled RFS date Q1
2005 with 1.28Tb/s design capacity.
9
Could there be some oversupply?

Atlantic: 19% of total capacity lit (2,338Gb). Of lit capacity
about 1,300Gb is sold.
 Pacific: 16% lit (1,043Gb)
 Intra-Asia : 3.5% lit (15,810 Gb design capacity)
 US-Latin America: 6% lit ( 5,166 GB design)
 Europe-Asia: 30 gig lit ; 120 gig design capacity
 Europe-Africa-Asia: 10 gig lit ; 130 gig design
Numbers; Telegeography 2003 Int’l bandwidth report
10
Congratulations Iceland !
FARICE : 40 gig at RFS 640 gig design RFS jan 1st 2004
Ready to participate in lambdaswitching!
11
Predictions and forecasts revisited
• What
will fill the capacity and how fast?
• Current (2003) transatlantic: voice: 9.3gig
internet: 258.3gig other (IPL etc): 48.7gig
•TeleGeography predicts a slow growth
scenario of 763.6gig and a fast growth of
1.48Tb for 2007
•Who would dare to predict it will take 17
years to fill the capacity?
•Could lambda switching have the same
disruptive effect on predictions and forecasts
as internet was about to have when planning
capacity a decade ago?
12
Who still remembers Icecan and Scotice?
Laid in 1961-62, capacity: 24
telephone channels
Also in 1961, COTC as
Teleglobe was known in those
days together with BPO and
C&W activated the first Cantat
between New Foundland and
Scotland with a capacity of 80
telephone channels. Cantat1
was retired in 1986
From the Bill Glover cable
stamp collection
See: http://www.atlanticcable.com/
13
Disruptive capacity growth?
It happened before
In the 1950s new technology put cables ahead of
radio. Small vacuum tubes that could operate
under water for 20 years or more meant that
amplifiers could be buried at sea with the cable.
This boosted the cable's information capacity to
the point that it could even carry telephone
signals.
Small vacuum tubes like this could be buried at
sea with the cable for years. They helped to
increase a cable's information-carrying capacity by
more than a thousandfold.
Borrowed from : The Underwater web, Smithsonian Institute
http://www.sil.si.edu/Exhibitions/Underwater-Web/uw-credits.htm 14
Agenda

Evolution of Transoceanic Internet Capacity Demand
and Supply

Aftermath of the Great Telecom Storm

The R&E World Sees More and More Light

Next (Light)wave of Opportunities
15
Aftermath of the perfect storm

More than 100 billion in default
 Huge write-offs
 Market valuation telecom sector down 1 trillion $
 500,000+ jobs lost at service providers and manufacturers
 Carrier capex still very conservative
 First wave emerges from bankruptcy protection
 Bottom reached but slow recovery
16
How did we get into this predicament?

Deregulation + internet and wireless boom + abundance of
equity capital --» wild spending by established telecom
carriers and start-ups.

1996 US Telecom Act and European deregulation
promised access to a US$300 billion market growing at
10% p.a.

Emulation of get rich quick model by 1996 purchase of
MFS by Worldcom for US$14billion or 6 times the value of
assets put in the ground

Spectacular advances in DWDM technology expected to
accomodate an insatiable bandwidth demand.
17
How did we get into this predicament? (2)

Unrealistic expectations of traffic growth
 Does internet traffic double every 90 days or every
year? depends on what scale you look at it.
 Rising multipolarity of the internet was largely
ignored in early models : end of the US centricity
of information
– Japan : 80% of accessed internet info is local.
– Chile: 70% is local
– USA: 10 to 30% of accessed information resides in the
region!
18
During the storm : progress continued

20+ million Broadband internet accesses (DSL and
cable) in NA by end of 2003

2 million personal Wi-Fi routers (Linksys, D-link etc)

The Wi-Fi hotspot phenomenon

Cellphones become multifunctional and start to replace
fixed line
Stage set for the next wave : global reachability
and mobility
19
Agenda

Evolution of Transoceanic Internet Capacity Demand
and Supply

Aftermath of the Great Telecom Storm

The R&E World Sees More and More Light

Next (Light)wave of Opportunities
20
The R&E world savours the bandwidth glut

Happy days for the R&E world:
– Europe: Geant goes 10 gig, some NREN’s also
– North-America: lambda’s and dark fibre
– Transatlantic: Finally enough to satisfy the
bandwidth gluttony of the high energy physics
people. 10gig transatlantic links on the verge of
becoming common place.
– Transpacific and intra-Asia: slower price decline,
lambda’s still have to wait a while.
– Europe-Asia : remains a bottle-neck however.
21
North-American R&E lambda initiatives

Canada
–
–
–
–

Ca*net4 : Canarie federal R&E network
RISQ : Quebec
ORANO : Ontario
BCnet ORAN: British Columbia
USA
–
–
–
–
–
–
NLR (National Lightrail): CENIC, Cisco, Level3
Fiberco : Internet2 with Level3
USAwave : SURA with AT&T
Teragrid
DoE ultrascale : initially ORNL – Sunnyvale - Chicago
Abilene 2nd gen
22
USA regional R&E initiatives













California (CENIC Optical Networking Initiative)
Connecticut (Connecticut Education Network)
Florida (Florida LambdaRail)
Indiana (I-LIGHT)
Illinois (I-WIRE)
Maryland, D.C. & northern Virginia (MAX)
Michigan
New York + New England states (NEREN)
North Carolina (NCNI)
Ohio (Third Frontier Network)
Oregon
SURA Crossroads (southeastern region)
Texas (Star of Texas)
Source: Paul Love
internet2 Jtech
Lawrence,Ka august 4th
23
Transatlantic R&E lambda initiatives

Translight
–
–
–
–

Starlight : Eurolink -UIC
Netherlight : SURFnet
DataTAG/CERN
Canarie
10 gig triangle Chicago-Amsterdam-Geneva and
Ca*net4 10 gig to NY and Seattle
24
The capacity divide

Uneven geographic distribution of capacity gluts
contributes to a capacity divide, sometimes further
exacerbated by monopolies or oligopolies in certain
regions

Clearly illustrated by the SLAC PingER project
measuring regional disparities of internet packet loss.

Abrupt halt of the global build affected Mediterranean
and Europe-Asia

Uneven capacity distribution is also visible on
national and regional level: Everyone builds on same
major routes and same major population centers.
25
Agenda

Evolution of Transoceanic Internet Capacity Demand
and Supply

Aftermath of the Great Telecom Storm

The R&E World Sees More and More Light

Next (Light)wave of Opportunities
26
The optical future has already started




Will it lead to an all-optical future?
Will Moore’s law and related laws for growth of fiber
transmission capacity and internet growth continue to
apply? Probably
The laws of gravity still apply, even in the New Economy.
Progress alternates between periods of exponential
growth and plateaus were the progress is absorbed.
Technology ahead of demand?
– 160 wavelengths at 40Gb ?
– Soliton technology?
– Optical crystals and hollow-core fibers with another
100fold increase of capacity per fiberstrand capacity?
27
The Verizon optical bet
As reported in Business Week, August 4th

Verizon plans fiber to every home and business in its 29
state territory : 10-15years and US$20 to 40 billion.
US$12.5 capex in2003.

Why? Cable Companies are eating into phone lines (2.2
million end 2002, forecast 3.7 million in 2005 ) and are
ahead in broadband internet (66% of the 18 million US
BB internet users).

Is this model applicable outside the US in coming years?
– Not sure; phone companies DSL dominate in many countries.
Competitive pressure from cablecos mostly not so severe.
28
What are the next telecom growth engines?

The telecom ecosystem is famished. Hopes for reviving
corporate and end-user demand are largely pinned on
–
–
–
–
–
–
–
–
–
–

Integrated mobile internet access (E-mail,web, data), 3G
SMS, Voice over IP, location based services
Home/SME area networks
Local wireless: Bluetooth, Wi-Fi, Ultra large Bandwidth
Further penetration of DSL and cable access, FTTH?
P2P applications : videoconferencing, gaming etc.
Secure VPN’s and end to end security and encryption.
Remote monitoring, tracking, sensing (healthcare, transportation etc.)
Audio/videostreaming
sensor networks, RFID
Widespread penetration of this end to end mobility and
reachability on the internet implies the deployment of IPv6,
prerequisite for permanent addresses, scaleability and
29
sufficient address space.
Next: the era of ubiquitous everything

Ubiquitous computing

Ubiquitous communication

Ubiquitous information access

Ubiquitous monitoring

Ubiquitous localisation and tracking

Ubiquitous neighbour discovery and sentient
networks
30
Thank you for your attention
31