Towards advanced CEF Networks lighting

1
Towards advanced CEF Networks lighting
Research activity Optical networks
www.ces.net
czechlight.cesnet.cz

• Stanislav íma, Lada Altmannová

2
Authors participate on Optical networks activity
of CESNET research intent, Joint Research
Activity of GN2 project, SEEFIRE project, Porta
Optica Study project, Lucifer project.

• Presented ideas are of authors
• and do not necessarily reflect an official
  opinion
• of the CESNET or any other institution or
  project,
• and are opened to improvement.

3
Why dark fibre for Research and Education
Networks (RENs)?(we have published this slide in
TNC2004, Rhodes)

• Freedom in REN design
• Fixed costs of long-term use
• Transmission capacity up to Tbps
• Cost-effective ways of sharing (WDM, TDM, )
• Multiple wavelenghts possible (e.g. 1 256)
• Management of transmission
• Equipment selection, moving, design, ..
• Now Move of Research and Education Networks to
  dark fibres is successful and continues, some
  other networks follow RENs (municipalities,
  hospitals, military, )

4
Why open lighting for Research and Education
Networks?(open means multi-vendor under rules
given by RENs)

• Freedom in REN design on optical and
  optoelectronic layers (L1L2)
• Save long-term investments on dark fibres (change
  lighting instead of fibres)
• Fibre is resource (facility) and sharing of
  fibres is the biggest opportunity to decrease the
  overall costs of networking.
• Use of equipment of many vendors for lighting and
  L1 interoperability is inevitable due to
  multi-domain nature of research networking
• Multi-vendor lighting in single domain is
  inevitable to take full advantage of fast
  development of photonic industry and brings new
  functions, better parameters and/or
  cost-effectiveness
• No vendor has the best lighting equipment for all
  needs
• Roadmaps of technology development of big vendors
  are not optimal to REN needs
• Some big vendors use obstructions to suppress
  interoperability

5
Multi-purpose dark fibre footprint for research

• Design in independent domains results in separate
  fibres for
• National RENs and Regional Optical Networks
• Continent-wide RENs (GEANT2, NLR)
• National dark fibre testbeds (CzechLight, VIOLA,
  )
• Prepared continent/world-wide dark fibre testbeds
  needed for research of networking (see for
  example GENI, FP7)
• Collaborative design by using open lighting is
  able to share fibres by means of WDM for all of
  above purposes, for example
• Fibres of NRENs connected by CBFs are shared by
  GEANT2 lambdas
• Lighting of fibres Brno-Bratislava and
  Brno-Vienna are prepared for sharing by IP
  service, e2e lambdas including GLIF, and testbed
  lambdas
• Lighting of fibres to end users are prepared for
  sharing by IP service, e2e lambdas and testbed
  lambdas
• This means dark fibre testbed (or facility
  network) implements lambda infrastructure for
  production network instead of buying lambda
  service! (no lighting devices disruption is
  allowed in such dark fibre testbed)

6
National dark fibre footprints for research
(draft)
7
European dark fibre footprint for research (draft)
8
Leased digital circuits for extension - example
9
Collaborative design of multidomain CEF networks

• Primary rule for CEF networks fibre topology is
  KIS (but now it means Keep It Short) fibre is
  the most expensive part of network (for example
  60)
• How to implement continental-wide services and
  testbeds by means of national fibre footprints
  and cross-border fibres and in what cases we need
  also parallel continental-wide fibre
  infrastructure?
• New types of transmitters, receivers, amplifiers,
  gratings etc. are available, some of them even
  with MultiSource Agreements (this is very
  important for designers), with better parameters
  and very cost-effective
• Overcome distances by lambdas only is possible
  saving number of routers, switches and
  administrators

10
Railways example

• Mesh network without parallel lines (with
  exceptions)
• Parallel railway lines are very expensive !
• Reasons for parallel lines
• Transport speed, local traffic
• Heavy workload
• Different owners, etc.
• Parallel lines are rare!
• No overlaying domains

11
Why open lighting(the first indication of
advantages for production network)

• 189 km Praha Pardubice , G. 652, 46 dB
• STM-16 since March 2001
• Lighting equipment 1x ONS15104, in-line, protocol
  dependent
• Total 67.300 Euro
• STM-16 or GE since May 2002
• Lighting equipment 2x OA Keopsys 2x 21 dBm, NIL,
  protocol independent
• Total 37.300 Euro
• Totals above are for varying parts of STM-16
  lighting, other parts are the same for both cases

12
Obr v RGB
13
Why open lighting(the second indication of
advantages for production network)

• 123 km Praha Plzen , G. 652, 33,7 dB
• STM-16 since September 2003
• Lighting equipment 1x ONS15104, in-line, protocol
  dependent
• Total 67.300 Euro
• STM-16 or GE since December 2003
• Lighting equipment 2x OA Keopsys 2x 10 dBm, NIL,
  one side amplification only, protocol independent
  
• Total 20.000 Euro
• Now we use CzechLight Amplifiers (CLA) with
  improved parameters, cost and Unix remote
  management

14
Why open lighting(Advantages for multi-purpose
dark fibre footprint)

• 190 km Brno Bratislava , G. 652, 45 dB
• n x 10 GE since February 2006
• Lighting equipment 2x
• CASE MUX/DMX FBG, CLA PB01, XENPAK
• Total 45.000 Euro
• This is cost for one 10G lambda with extension
  possibility up to 4 lambdas, adding of one lambda
  costs 10.000 EUR (two XENPAKs). If you need
  extensibility to 8 lambdas, you can use more
  expensive MUX/DMX (add 1.260 EUR for one line).
  If you need extensibility to 16 lambdas, you can
  use OAs with a little higher output power. The
  same XENPAK from other vendor costs 20.000 EUR/pc
  (third party components are not allowed). If you
  can use XFP instead of XENPAK, price will be
  about 3.800 EUR/pc.
• Prepared 224 km Brno - Vienna (ACONET, VIX)

15
Multi-purpose using of dark fibre
NREN PoP 1
NREN PoP 2
End User Y
End User X
End User 1
End User 2
IP service
IP service
Testbed
Testbed
e2e service 1
e2e service 1
. . .
. . .
First mile fibre
Core fibre
First mile fibre
e2e service n
e2e service n
DWDM MUX/DEMUX
DWDM MUX/DEMUX
CLA PB01
Lambda for production IP service
Experimental lambda for testbed
NREN PoP n
Planned lambda for e2e service
16
(No Transcript)
17
CzechLight December 2005
18
Multi-service connections of hospitals
Praha PoP
Ústí n.L. PoP
End User Y
End User X
Central Military Hospital Prague (CMH)
Masaryk Hospital Ústí n.L. (MHÚL)
CESNET2
CESNET2
CzechLight
CzechLight
CMH
MHÚL
. . .
. . .
130,6 km 29,7 dB
8,8 km 3,8 dB
1 km 0,3 dB
e2e service n
e2e service n
DWDM MUX/DEMUX
DWDM MUX/DEMUX
CLA PB01
Lambda for production IP service
Experimental lambda for testbed
Planned lambda for e2e service
19
Acknowledgement

• Jan Gruntorád, Miroslav Karásek, Martin Míchal,
  Václav Novák, Jan Radil, Karel Slavícek, Radek
  Velc, and Josef Vojtech from CESNET
• All partners from CEF Networks community