12/08/01Ernst1 Recent Development in Networking Connectivity Michael Ernst DESY.

12/08/01 Ernst 1

Recent Development in Networking Connectivity

Michael Ernst

DESY

12/08/01 Ernst 2

DFN’s Trans Atlantic Connectivity Today

STM 4

STM 4

12/08/01 Ernst 3

News on TA Connectivity (1)

• Planned for 2002– Direct link GEANT <-> Abilene / CANARIE

@ 2 * 2.5 Gbps (Q I, 2002)– UCAID will add another 2 * 2.5 Gbps– UCAID proposing joint Project GTRN

12/08/01 Ernst 4

DFN’s Trans Atlantic Connectivity in Q I, 2002

STM 4

STM 16

STM 16

12/08/01 Ernst 5

News on TA Connectivity (2)• Global Terabit Research Network

– “Terabit Net” for Science and Research– Europe as Mediator for Asian-Pacific Area

• For DFN/US Global Transit (commodity traffic)– 2 * 2.5 Gbps in Q I, 2002– Contract with 2 Providers

• Global Crossing (2.5 Gbps)• KPNQwest (2.5 Gbps)

– Tendering Process in Collaboration with DANTE

12/08/01 Ernst 6

Entwicklung US-Verkehr

0

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Sept '00 bis Nov '01

DE -> US

US -> DE

Trans Atlantic Traffic Development from Sep ‘00 - Nov ‘01 (Statistics provided by DFN)

12/08/01 Ernst 7

GÉANT

• Backbone speeds: – initial target: core at 2.5 Gbps– network achieved:

9 trunks at 10 Gbps and 11 trunks at 2.5 Gbps

• Access speeds:– 11 NRENs to connect at 2.5 Gbps

• Future plans:– 100’s of Gbps within four years

Gigabit Speeds

12/08/01 Ernst 8

GÉANT

• TEN-155:– 21 NRENs linked– 25 countries connected in total

• GÉANT:– 27 NRENS– 31 countries - 6 new countries, 2 in the Balkans– Open to further connections from national organisations

Geographic Expansion

12/08/01 Ernst 9

GÉANT

• Connectivity to other regions of the world

• European Distributed Access: – effective distribution mechanism through backbone core– uniform presentation of the European networks

Global Connectivity

12/08/01 Ernst 10

GÉANT

• TEN-155 continuity: – TEN-155 Managed Bandwidth Service, using ATM

technology

• End-to-end QoS– with NREN involvement

• Different types of QoS:– Guaranteed bandwidth– Predictable delay and jitter– Guaranteed bandwidth + predictable delay and jitter

Guaranteed QoS

12/08/01 Ernst 11

GÉANT - A Development Platform

• A network for research: – TF-NGN, SEQUIN

• http://www.dante.net/tf-ngn• http://www.dante.net/sequin

• Initiatives with IPv6: – TF-NGN, 6NET Proposal

• Co-operation with Grids: – EuroGrid, DataGrid

• Joint development clauses with suppliers

12/08/01 Ernst 12

Price Development of International Bandwidth (Source: V. Berkhout)

200,000 150,000

30,000

12,500

5,000

200,000100,000

20,000

2,000

36

1

10

100

1,000

10,000

100,000

1,000,000

1996 1997 1998-1999 1999-2000 2000-2001

Period covered

Eu

ro /

Mb

ps

/ Yea

r

average offer price

lowest offer price

12/08/01 Ernst 13

GÉANT Technology

• 2.5/10 Gbps as single wavelengths – with SDH framing– no access to optical level / (dark) fibre

• TEN-155 aggregate circuit length = 22,000 km

12/08/01 Ernst 14

Stepping to 10Gbps

• Cost vs capacity– 10Gbps in 8 countries …

• quadruples capacity• adds less than 10% to cost

– Pushing 2.5Gbps as far as possible

12/08/01 Ernst 15

Issues

• Service is diverse (wavelengths and SDH)• Guaranteed bandwidth and VPN capability

needs further development . . . • . . . as does management of end-to-end

capability• Cooperation of NREN, MAN and campus

network operators is needed

12/08/01 Ernst 16

Summary on GEANT

• 10 Gbps now!

• Wavelength networks– National and International

• Better geographical coverage

• Progress on global connectivity

• Challenging the vendors

• QoS/VPN will need our help!

12/08/01 Ernst 17

STM 16(2.5 Gbps)

DFN’s Connectivity to Europe

UK

BE

NL

ITCH

DEFR

SE

AT

LV

PL

SK

CZ

HR

HU

EE LT

LU

622

622

SI

RO

PT

ES

IE

BG

CY

GR

622

34

34

45

155

155

155

155

155

155

34 155

34

155 34 45

10G

2.5G

O ther

IL

GEANT

12/08/01 Ernst 18

Entwicklung TEN-155/GEANT - Verkehr

0

5

10

15

20

25

30

35

40

45

50

1 2 3 4 5 6 7 8 9 10 11

Jan '00 bis Nov '01

TB

yte DE -> EU

EU -> DE

Development of DFN <-> TEN-155 / GEANT Traffic from Jan ‘01 - Nov ‘01 (Statistics provided by DFN)

12/08/01 Ernst 19

0500

100015002000250030003500400045005000Ja

n

Mar

May Ju

l

Sep

Nov

Total

Internat'l

GB

DESY’s Monthly Traffic Volume (Received)

12/08/01 Ernst 20

The Silk Project

Hans Frese DESY Hamburg

NATO Advisory Panel on Computer Networking

SILK Task Force

12/08/01 Ernst 21

The Silk Task Force

• Peter Kirstein, UCL, London• Hans Frese, DESY, Hamburg• Robert Janz, RUG, Groningen• Sergey Berezhnev, MSU, Moscow• Zita Wenzel, ISI, Marina del Rey, CA• Rolf Nordhagen, UIO, Oslo• Ruben Mkrtchyan, YERPHI, Yerevan• Walter Kaffenberger, NATO

12/08/01 Ernst 22

Activity of the Network Panel

• The Networks Panel has supported Network Infrastructure Grants (NIGs) for many years

– Was initially Russia and Eastern Europe

– Southern Caucasus and Central Asia are current principal areas for our larger grants

• Internet Connectivity has been a large part of each NIG

• Current bandwidths much too small

– but all that can be afforded from budget

12/08/01 Ernst 23

Intentions of NIGs

• Improve National Research Net Infrastructure– Not that of isolated groups or institutes

• Encourage National Collaboration– Preferably to set up National Research and

Education Networks (NRENs)

• Encourage International Collaboration– Ever more important at the current time

12/08/01 Ernst 24

Current Connectivity

• Bandwidth from NATO sources currently 64 – 512 Kbps– Would like to go up by an order of magnitude at

least– Cost unaffordable in current model

($100k per year for 1 Megabit per second)

• National Research and Education Networks (NRENs) partially exist in most of the countries intended currently

12/08/01 Ernst 25

Possible Technologies

• Mainly Fibre in Western Europe– No affordable fibre yet in Caucasus or

Central Asia (> 5 times satellite cost)– Does exist in E. Europe and Russia

• Satellite attractive in these areas– Satellite Bandwidth driving force– Broadcast capability can be useful

• Proposed Silk Project in 2000 – Based on VSAT Technology

12/08/01 Ernst 26

Schematic of the Silk System

12/08/01 Ernst 27

A short primer on satellites (1)

• Satellites are bent pipes in the sky• 5 to 10 year lead times imply mature/old

technology in the sky• Fiber has taken over the oceans, satellites are

looking for work• Compared to fiber, satellite bandwidth is low,

but ...

12/08/01 Ernst 28

A short primer on satellites (2)

• ... one satellite covers one third of the globe• ... broadcasting to many locations is trivial• ... bandwidth is simplex

– it can be allocated asymmetrically and shared between locations

• ... minimal local infrastructure requirements– unobstructed view south– 240 Volts with diesel backup if needed

12/08/01 Ernst 29

A short primer on satellites (3)

• Buying satellite bandwidth:– you pay for radio frequency bandwidth– and battery consumption in the sky– buy in bulk to obtain discount

• Using a larger dish fetches more energy and improves the signal to noise ratio

• This allows higher density modulation which produces more Mbps per MHz

• Net result: 1 Mbps per year for $25K

12/08/01 Ernst 30

Who gets connected?

• Funded by NATO– National Research and Education Networks

(NRENs) in the Partner countries

• Co-funded by– NGOs– Supranational Organisations

• Staged Implementation– E.g. Armenia, Georgia, Kyrgyz Rep, Uzbekistan– E. Azerbaijan, Kazakhstan, Tajikistan,

Turkmenistan

31Ernst12/08/01

Satellite station configuration

HUB

Satellite router

NREN router

Satellite dish

HSSI

100Mbps

12/08/01 Ernst 32

Planned Silk BandwidthSilk total bandwidth requirement

Per half year

0

5

10

15

20

25

30

01/H2 02/H1 02/H2 03/H1 03/H2 04/H1 04/H2

Total bandw idth inMbps

12/08/01 Ernst 33

EurasiaSat

• PROs– Covers both Asia and

Western Europe

– Can use small earth stations (2.4m)

– Hamburg hub (DESY) well connected to European backbone

– Bandwidth on demand for e.g. teleteaching

• CONs– Requires expensive

hub station in the West ($650K)

12/08/01 Ernst 34

Project Management

• DESY will provide Technical Management• Area Consultants plus Silk Task Force will

provide first Process Management– If project grows, may get professionals– May be able to work with other funding agencies

operating in the area

• Project Steering Committee provide high level policy management

12/08/01 Ernst 35

Policy Steering Committee

• Will include at least all Co-Directors and representatives of funders

• Terms of Reference to be decided by members

• Policy includes many areas to discuss – AUP rules, Membership, responsibilities of

NRENs, move to financial sustainability, bandwidth rules, any dispute resolution

12/08/01 Ernst 36

Summary

• Propose a system with 25-50 Mbps for 8 countries

in Caucasus and Central Asia

• System could grow with additional investment from others

12/08/01 Ernst 37

Conclusions

• The Project is ready to start

• The technical and organisational framework

is in place

• A choice has to be made

12/08/01 Ernst 38

L3 Switch = Router

Backbone

Server Farm

ServerDistribution

L3 Switch = Router

L2 Switch

Modular Campus Network DesignBuilding

AccessLayer

DistributionLayer

Building BuildingL2 Switch

Backbone