Download 1/4 Qualificação como Organização Social - lishep

Advanced networking in Latin
America and the CLARA initiative
LISHEP 2004
Rio de Janeiro, Brazil
February 2004
Michael Stanton
CLARA Technical Committee
Rede Nacional de Ensino e Pesquisa do Brasil - RNP
[email protected]
A Brief Story of Networking in Latin America
• Political, linguistic and cultural considerations have
traditionally led to considerable interaction between
countries within the region
However, networking has not followed this model:
• First connections (BITNET) starting 1986 using satellite
links between the US and each country separately
• Same topology inherited with transition to Internet
• Even multilateral initiatives (RedHUCyT in mid 90s and
AMPATH from 2001) have used traffic hubs in the US.
Michael Stanton - LISHEP 2004
2
First global conections from LA countries
Two “classical” phases of
connectivity:
• e-mail networks (BITNET,
UUCP)
• full Internet (IP) connectivity
• Table shows the dates of
the first connections for
each LA NREN (National
Research and Education
Network)
MX CL BR NI
UY PY VE AR CR
e-mail
86
86
88
88
88
89
90
90
90
IP
89
92
91
94
94
95
92
93
93
CO EC PE BO CU PA GT SV HN
e-mail
90
91
91
91
91
92
92
94
94
IP
94
92
94
95
96
94
95
96
95
Michael Stanton - LISHEP 2004
3
Michael Stanton - LISHEP 2004
4
Influence of telecommunications infrastructure
• Until very recently, the only available telecom
infrastructure for data communication was by satellite
– cost independent of distance
– no incentive for establishing links within the region, as
all countries were mainly interested in access to
global Internet
• Recent important changes (since late 1990s):
– end of state telecom monopoly in many countries
• competition and lower prices
• most LA NRENs replaced by commodity IP
providers (for economic or political reasons)
– building out of new infrastructure based on submarine
fibre optical cables
Michael Stanton - LISHEP 2004
5
Optical cable infra-structure
• Advances in optical transmission technologies have
recently made it possible to build very long distance
undersea communications systems based on DWDM
• In the late 1990s, several new DWDM (Dense
Wavelength Division Multiplexing) cable systems were
built, vastly increasing the installed capacity
• Principal new undersea cable operators in Latin America:
– Global Crossing
– Telecom Italia – Sparkle
– Telefonica International Wholesale Services - TIWS
(E-mergia)
– New World Networks (ARCOS cable)
Michael Stanton - LISHEP 2004
6
New Optical Cables in Latin America
to
New York
to
California
Miami
San Juan, Puerto Rico
E-mergia (TIWS)
Global Crossing & TI Sparkle
Global Crossing
ImpSat
Transandino
UniSur
Michael Stanton - LISHEP 2004
7
New cables in the Caribbean (Maya & Arcos)
North Miami
309km
474km
271km Cat Island
319km
521km
Maya
Crooked Island
Cancun
258km
165km
Providenciales
(Turks & Caicos Islands)
376km
Tulum
325km
363km
Ladyville
Arcos
Puerto Plata
291km
(festoon)
San Juan
Punta Cana
294km 241km Trujillo
Puerto
Puerto 339km
Barrios
114km Cortes
Puerto
Lempira
258km
1006km
Puerto
Cabezas
279km
372km
Bluefields
351km
270km
371km
Puerto
Limon
301km
Maria
Chiquita
242km Curacao
Willemstad
Punto Fijo
Riohacha
314km
Ustupo
Michael Stanton - LISHEP 2004
8
Present Advanced R&E Connectivity
in Latin America
AmPath
• uses Global Crossing
• connects AR, BR (2), CL,
VE
• 45 Mbps (one size fits all)
• all connections are point to
point from Miami, and
thence to Abilene
AmPath
Mexico
• cross-border connections to
USA (TX and CA)
Michael Stanton - LISHEP 2004
9
Present State of Latin American NRENs
Established education and research networks:
• With dedicated Advanced R&E connections:
Argentina, Brazil, Chile, Mexico, Venezuela
• Some with dedicated int’l connectivity:
Costa Rica, Cuba, Uruguay
Education and research networks being re-established
(present nat’l/int’l connectivity through commercial ISPs)
• Bolivia, Colombia, Ecuador, Guatemala, Panama, Peru, Paraguay,
El Salvador
No education/research network (most connected to Internet via
commercial ISPs):
• Nicaragua, Honduras, Dominican Republic, Haiti, rest of Caribbean
Michael Stanton - LISHEP 2004
10
Argentina - RETINA (www.retina.ar)
-
45 Mbps to
AmPath
Abundant
Medium
Narrow
•4 with advanced connectivity
•8 in the near future
•57 with low connectivity
Michael Stanton - LISHEP 2004
11
Brazil - RNP (www.rnp.br/en)
-
-
-
ATM backbone
- 14 nodes
- 300 Mbps total
b/w
FR to other PoPs
15 state networks
Aggregate int’l b/w
over 500 Mbps (incl.
90 Mbps to AmPath)
new RNP backbone in
2004 – 1.8 Gbps total
b/w (6x increase)
Michael Stanton - LISHEP 2004
12
Brazil – RNP: probable 2004 backbone topology
TO
PI
AC
SE
AL
MG
CE
MA
RO
RN
MT
PE
DF
MS
PB
GO
BA
RJ
SP
RR
STM-4 (622 Mbps)
AP
ES
PR
AM
PA
RS
SC
Pop already
tendered
Future tender
Michael Stanton - LISHEP 2004
STM-1 (155 Mbps)
E3 (34 Mbps)
E1 (2 Mbps)
13
Chile - REUNA (www.reuna.cl)
Arica
- ATM
backbone
- 10
nodes
- 10/60
Mbps
- 45 Mbps to
AmPath
Iquique
Iquique
uta
unap
Antofagasta
ucn
uantof
Antofagasta
Copiapó
uda uda
Copiapó
La Serena
userena
La Serena
Valparaíso
Valparaíso
utfsm
Santiago
Santiago
uchile
Talca
utem
utalca
Talca
umce
Concepción
udec
Con cepción
Valdivia
reuna
Temu
co
Osorno
udp
Temuco
ubiobio
ufro
Valdivia
uach
unap
uta
ucn
uantof
uda
userena
utfsm
uchile
utem
umce
udp
utalca
udec
ubiobio
ufro
uach
ulagos
ulagos
Universidad Arturo Prat
Universidad de Tarapacá
Universidad Católica del Norte
Universidad de Antofagasta
Universidad de Atacama
Universidad de la Serena
Universidad Técnico Federico Santa María
Universidad de Chile
Universidad Tecnológica Metropolitana
Universidad Metropolitana de Cs. de la Educación
Universidad Diego Portales
Universidad de Talca
Universidad de Concepción
Universidad del BíoBío
Universidad de la Frontera
Universidad Austral de Chile
Universidad de los Lagos
Switch de Backbone
Switch de Acceso
Router de Acceso
Michael Stanton - LISHEP 2004
Trama SDH
FO Multimodo
FO Monomodo
14
Mexico - CUDI (www.cudi.edu.mx)
• Internal links at 155
Mbps
• 400 Mbps of int’l
connectivity
Michael Stanton - LISHEP 2004
15
Where do we go from here?
• AMPATH´s achievements
– Initial boost for Advanced Networking in LA
– Stimulus for advanced connectivity inside each country
– Motivation for collaborative projects
– Connectivity needs, delayed till now due to high costs,
being solved
BUT
• Why does LA communicate internally through Miami?
• Why does LA communicate with other parts of the world
through the US?
Michael Stanton - LISHEP 2004
16
Another paradigm: Pan-European R&E networking
• In Europe, global networking also began with direct
BITNET and IP links to the US from separate countries
• Since the early 1990s great efforts have been invested in
pan-European networking, through the creation of a
series of regional backbone networks:
• These networks have been built and managed by
DANTE (Delivering Advanced Networking Technology to
Europe), with financing by European NRENs and the EU
• Four versions of the pan-European backbone network
– EuropaNET (1992-1997)
– TEN-34 (1997-1998)
– TEN-155 (1998-2001)
– GÉANT (2002- )
Michael Stanton - LISHEP 2004
17
TEN-34
• Trans-European
Network at 34
Mbps
• 20 countries
• operational in
1997
• backbone speed
inferior to internal
NREN links
(cost of int’l links)
Michael Stanton - LISHEP 2004
18
TEN-155
• Set up after
liberalisation and
harmonisation of
European telecom
industry
• Much cheaper int’l
connectivity within
Europe
• In some countries
liberalisation delayed
Michael Stanton - LISHEP 2004
19
GÉANT
• First network of the
“Bandwidth Age”
• 20-fold increase in
capacity over TEN-155
for the same cost
• Principal connections
are 10 and 2.5 Gbps
wavelengths
• Currently the largest
capacity operational IP
network in the world
Michael Stanton - LISHEP 2004
20
The @LIS iniciative
• Through GÉANT, the European R&E community enjoys
high bandwidth connectivity with N. America
• Initiatives already taken to improve connectivity to AsianPacific, Mediterranean and Latin American regions, with
support from the European Commission
• @LIS: Alliance for the Information Society (2003-2005)
– 62.5 Million Euros for EU-LA on Information Society
Issues
– 10 Million Euros for Interconnecting Europe & Latin
American Researchers
Michael Stanton - LISHEP 2004
21
CAESAR: Connecting All European and
South American Researchers.
European initiative to prepare for the @LIS program
• Promote EU-LA connectivity through regional connectivity within
LA plus a large pipe to Europe
• Participants: DANTE, NRENs of Spain and Portugal
• CAESAR Workshop 2002 in Toledo became starting point for
CLARA
– cooperative organisation for advanced networking in LA
– regional network:
feasibility study showed that @LIS budget sufficient to
establish advanced connectivity to all LA countries
• EU has agreed to 80% - 20% cost sharing between EU and LA
Michael Stanton - LISHEP 2004
22
• Association of NRENs in LA open to all LA Countries
– constituted in Uruguay (like LACNIC)
• CLARA is not limited to @LIS/CAESAR time scale and
restrictions
• Will connect LA to Europe and to other regions
• Cost to connect to the backbone will be the same for every
country at equal bandwidth
• Improve Internet2 connectivity by optimising LA
participation in AMPATH
Michael Stanton - LISHEP 2004
23
CLARA Members
•
•
•
•
•
•
•
•
•
•
Argentina
Brasil
Bolivia (*)
Chile
Colombia (*)
Costa Rica
Cuba (*)
Dominican Republic (*)
Ecuador
El Salvador
•
•
•
•
•
•
•
•
•
Guatemala
Honduras (*)
Mexico
Nicaragua
Panama
Paraguay
Peru
Uruguay
Venezuela
(*) expected future member
Michael Stanton - LISHEP 2004
24
Comments about CLARA
• CLARA responds to long-standing need for coordination
between LA NRENs.
• Builds on trust-building already carried out between
major partners
• Offers support for NREN building in other LA countries by
provision of support and int’l connectivity
Michael Stanton - LISHEP 2004
26
ALICE project: May 2003 to April 2006
ALICE - América Latina Interconectada Con Europa (2003-2006)
• Successor project to CAESAR
• Coordinated by DANTE, with participation of NRENs from Italy,
France, Spain, Portugal and the CLARA countries, and eventually
CLARA itself
• February 2003: technical definitions complete
• June 2003: Open tender for provisioning of links
• March 2004: Link contracts assigned
• May 2004: Network operational
Notes:
• DANTE is the project coordinator and will sign contracts with users
and providers
• CLARA is expected to represent interests of LA users in the
medium term (one year)
Michael Stanton - LISHEP 2004
27
Suggested network topology (tender document)
•
•
•
Major connectivity between
Argentina, Brazil, Chile and
Mexico (at least 45 Mbps)
Other countries connect to
major nodes (between 10
and 45 Mbps)
Large pipe to Europe (at
least 155 Mbps)
Michael Stanton - LISHEP 2004
28
Probable topology of CLARA network
Operator A
Operator B
Operator C
Operator D
Operator E
Bandwidth
characteristics:
•155 Mbps backbone
ring
•622 Mbps to Europe
•10 to 45 Mbps spur
links
•satellite link to Cuba
Michael Stanton - LISHEP 2004
29
The Digital Divide in Latin America – the cost
of access circuits to the CLARA network
•The proposed costs of access connections from NRENs
to the CLARA backbone were initially extremely high in
the case of Central America and Colombia
•This has required that these countries’ NRENs
negotiate better pricing with their national provider
Annual cost for access
circuits from CLARA
backbone to country point
of entry
Internal country costs for
access circuits
2 Mbps
€ 347.780
€ 1.051.077
10 Mbps
€ 982.033
€ 5.061.526
34 Mbps
€ 3.310.757
€ 9.932.272
Michael Stanton - LISHEP 2004
30
Final comments on the CLARA network
•
The network will provide connectivity for R&E collaboration traffic (i.e.
Internet 2 traffic) within the region and to other regions
– Initially connected to GÉANT (Europe)
– Expected soon to be connected via Tijuana (MX) to US networks
via Los Angeles peering point (agreement with CENIC)
• this also will provide access to APAN (Pacific Rim), as well as
other connected R&E networks
•
Initial backbone ring bandwidth of 155 Mbps
(connecting BR-AR-CL-PA-MX)
Initial connection to Europe at 622 Mbps from São Paulo
– RNP will use 155 Mbps of access capacity initially
CLARA network expected to be upgraded soon to support
international scientific collaborations between US and South America
•
•
Michael Stanton - LISHEP 2004
31
Experimental networking in Latin America
• Advanced R&E networks provide for today’s connectivity
needs of the R&E community
• In future, with the growth of this community and its needs,
new solutions will need to be provided
• Such solutions are currently being developed and
demonstrated in experimental networking testbeds
• NSF 2002 classification of networking testbeds (beyond
Internet2)
– Experimental Infrastructure Networks (EIN)
• provides stable networking infrastructure for
application development and demonstration
– Networking Research Testbeds (NRT)
• permits development of networking technology
Michael Stanton - LISHEP 2004
32
Some current experimental optical
networking projects in Latin America
• Chile:
G-REUNA - Advanced Applications Testbed
• Brazil:
Project GIGA - Optical Networking and Applications
Testbed
Both of these are a mixture of EIN and NRT
Michael Stanton - LISHEP 2004
33
G-REUNA (Chile)
experimental network (2002-2003)
• Phase I of G-REUNA:
• R&D in optical
networking and
advanced applications
• IP/DWDM
• govt. and telco support
• 250 km network between
Santiago and Valparaiso
• participation of leading
research universities and
national academic
network (REUNA)
• http://redesopticas.reuna.cl
Michael Stanton - LISHEP 2004
34
Project GIGA (Brazil)
Partnership between
– RNP (Brazilian NREN) www.rnp.br
– CPqD (telco industry R&D centre in Campinas, SP)
www.cpqd.com.br
– R&D community in industry and universities
• Build an advanced networking laboratory (GIGA network)
for development and demonstration purposes
• Support R&D subprojects in optical and IP networking
technology and advanced applications and services
• Industry participation (telcos provide the fibres; technology
transfer of products and services required)
• Government funding for 3 years - started December 2002
Michael Stanton - LISHEP 2004
35
GIGA network: objectives
• explore user control of optical fibre infrastructure
– interconnect 20 academic R&D centres in S.E. Brazil
– use of IP/DWDM with Ethernet framing
• provide Networking Research Testbed (NRT) for optical
and IP network development
• provide Experimental Infrastructure Network (EIN) for
development and demonstration of applications in
several research areas
• expected to operate in April 2004.
Michael Stanton - LISHEP 2004
36
GIGA network: geographical localisation
(states of São Paulo and Rio de Janeiro)
Universities
IME
PUC-Rio
UERJ
UFF
UFRJ
Unicamp
UNIFESP
USP
R&D Centres
CBPF
- physics
CPqD
- telecom
CPTEC - meteorology
CTA
- aerospace
Fiocruz - health
IMPA
- mathematics
INPE
- space sciences
LNCC
- HPC
LNLS
- physics
About 600 km extension - not to scale
LNCC
CTA
INPE
CPqD
LNLS
Unicamp
CPTEC
telcos
UNIFESP
USP - Incor
USP C.Univ.
Michael Stanton - LISHEP 2004
UFF
CBPF
LNCC
Fiocruz
IME
IMPA-RNP
PUC-Rio
telcos
UERJ
UFRJ
37
GIGA Project: Initial design of the network
• DWDM WAN between Campinas and Rio de Janeiro
Campinas
• WDM MANs in Rio,
S. Paulo and Campinas
• Switches between WAN
São
Rio de
Paulo
Janeiro
and MANs for IP packets
and lambdas (under study)
• later: redundant topology
and optical switching
S.J. dos
C. Paulista
Campos
São Paulo
S. José dos
Campos
Cachoeira
Paulista
MAN
SP
Rio de
Janeiro
Campinas
Petrópolis
Niterói
MAN
CP
MAN
RJ
Michael Stanton - LISHEP 2004
38
Some GIGA R&D subprojects
•
•
Networking technologies
– intelligent optical network with monitoring and control of physical
parameters
• optical amplification, dispersion, equalisation, SNR, ...
– optical switching architecture
• control plane: dynamical bandwidth provisioning and mesh
restoration
• provisioning end-to-end optical circuits for specific applications
– IP over WDM: unified control plane and integrated network
management
Applications:
– high performance distributed applications, including in HEP,
astrophysics, meteorology, health sciences, engineering,
biodiversity, etc.
– advanced multimedia applications
Michael Stanton - LISHEP 2004
39
What comes next in Latin America?
•
•
•
For RNP and REUNA, experimental networking projects are
important for two reasons:
– it provides a testbed network for R&D
– provides experience in setting up and running a “facilitiesbased network”, instead of one based on service offerings
from telcos.
• only need the raw physical medium, or perhaps access
to lambdas (wavelengths)
• perhaps the only feasible way to build and use really
high capacity networks
Next step for RNP and REUNA is to transfer this experience to
their production networks
– Many well-known examples from other countries
A logical consequence is to seek Gbps or lambda networking
services for international collaboration.
Michael Stanton - LISHEP 2004
40
Conclusion
• Both international cooperation (through AmPath and
CLARA) and development of experimental networking
have percussions:
– they provide valuable opportunities for academic user
community in LA to collaborate with peer groups in
other countries
– they permit the acquiring and diffusion of experience
in advanced networking technologies, often absent in
LA countries
– they lead to the provision of high capacity
infrastructure networks for scientific cooperation
Michael Stanton - LISHEP 2004
41
Acknowledgements and references
•
With thanks to many colleagues from both Europe and Latin
America, too many all to be mentioned here individually. Some
of the LA maps are by Florencio Utreras, from REUNA (Chile).
Material on European networks provided by Cathrin Stöver from
DANTE.
•
ALICE website:
www.dante.net/server/show/nav.009
•
ALICE brochure (in English, Spanish and Portuguese):
www.dante.net/alice/ALICEbrochure.pdf
•
RNP and REUNA websites:
www.rnp.br/en
www.reuna.cl
Michael Stanton - LISHEP 2004
42