Strategies for Metro/Regional Optical Networks

1
Strategies for Metro/Regional Optical Networks

• Brian Pratt, brian.pratt_at_meriton.com CEF/REN
  Conference, Prague, 18 May 2005

2
Agenda

• Overview of Meriton Networks
• Trends in Optical Networking
• Emerging research education applications and
  high-speed networks
• Technology
• Requirements for research education networks
• The current state-of-the-art
• Building Real High-Speed Optical Networks
• Choices for technologies
• Optical link engineering

3
About Meriton Networks

• Carrier-class, wavelength networking solution
• Enable the Growth of High-speed Metro/Regional
  Services, from Wall Street to Main Street
• Experienced team Leaders with Newbridge, Nortel
• Customers Growing base of enterprise and
  service provider customers
• Member of Internet2 HOPI Corporate Advisory
• Research Participant in CANARIE OBGP/UCLP
• Partnerships Fujitsu, Siemens, local partners
• Global Reach

46020 Network Management Platform
36170 MainstreetXpress
Corporate Headquarters Ottawa, Canada
USA Headquarters Raleigh, North Carolina
European Headquarters Bristol, UK
Asia Headquarters Hong Kong
4
Trends in Research Education Networks

• Large-scale applications outpacing network
  capacity
• Grids, 3D visualization, physics, astronomy all
  kinds of science, engineering and other research
  applications
• Non-wire-speed 10GigE is increasingly
  insufficient
• Multiple 10GigE wavelengths, moving to 40G
  deployment and research on 100G
• Some e-science applications already consuming 7
  Gbps of sustained bandwidth
• Many traditional/incumbent carriers not offering
  services or solutions to support these
  applications
• Metro/regional Optical Networks for research
  being built as an increasing rate
• Acquisition of dark fibre, lit up as private
  networks
• Different communities
• University/research focused
• Joint education/government initiatives
• Corporate/enterprise networks

5
Meritons Vision of theHybrid IP/Optical
Transport Network
IP/MPLS Network Layer
IP/MPLSRouter
IP/MPLSRouter
IP/MPLSRouter
IP/MPLSRouter
Least costly interfaces (e.g. 10GigE LAN
PHY, 1310 nm singlemode short-reach optics)
ServiceAccess Nodes
WXC
CWDM
Transponder function at the WDM layer
DWDM
DWDM
WXC
WXC
Multi-Service WavelengthTransport Network
High-EndUser
Metro
Access
Regional/Core
6
Status Quo SONET/SDH Transport Convergence
Multiple networking layers leads to additional
cost and operational complexities.
7
Todays Simplified Network
TDM
SAN
Ethernet
IP/MPLS
SONETSDH
Wavelengths
Waves
Fiber
Fiber
Converging to a full IP/MPLS layer over a common
wavelength network greatly simplifies the network.
8
Optical Networking in 2005
Technologies from the hype/bubble era of the late
1990s-2001are finally emerging as practical,
cost-effective solutions.

• All optical switching optical ROADM
• 1st generation wavelength blocker technology
  giving way to 2nd generation wavelength
  selectable switch (WSS) technology
• Electronic ROADM the benefits of wavelength
  switching simplicity of working in the
  electronic domain (OEO)
• Roadmap toward multi-degree optical ROADM
• Evidence of Moores Law applied to optical
  components
• Pluggable transceivers GBICs giving way to
  SFPs/XFPs
• 10G DWDM long-reach XFP transceivers for 8,000
  (was 200,000 5 years ago!)
• Multiple GigE wavelengths over regional distances
  now inexpensive
• Multiple 10GigE wavelengths now measured in
  x00,000, not x0,000,000!
• Some carriers offering wavelength services over
  shared infrastructure

9
Requirements for Optical Networks for RE

• Transparency
• Carry any service (Ethernet, SONET, SAN, etc.) at
  wire-speed, from GigE to 10GigE and beyond
• Ability to easily grow capacity to support 320
  Gbps (32?x10G) per fiber pair or more
• Support of emerging 40G and 100G wavelength
  technologies
• Ability to carry 10GigE LAN PHY natively
• DWDM equipment that interfaces to
  existing/most-cost-effective interfaces of
  GigE/10GigE routers/switches
• Avoid the expense and complication of 10GigE WAN
  PHY or SONET encapsulation
• Ability to carry alien wavelengths
• Support a mix of DWDM and CWDM, and interoperate
  between them
• Plug-and-play e.g. auto-discovery of new
  nodes/cards/interfaces
• Equipment that a technician can take out of the
  box and have up-and-running in hours, not days
• Simplicity must be like managing a router
  network
• Central management of all optical network
  elements, i.e. amplifiers, dispersion
  compensators, etc.
• No on-site visits to POPs required except to
  connect new users/fibers
• Similar management features as IP networks
  RADIUS authentication, packet counters, etc.
• Option for either in-band or out-of-band
  management or both
• Good tools for troubleshooting both CWDM and DWDM
  technology
• Hassle-free access to vendor expertise in optical
  network design and support

10
Requirements for Optical Networks for RE

• Simple and cost-effective, but with carrier-class
  reliability as/when required
• Use of pluggable transceiver technology, e.g.
  SFPs for lt 2.7G, XFPs for 10G
• Reduce costs, easy sparing, a standard with
  multiple sources
• Minimize optical loss
• High quality optical components that allow as
  many huts to be skipped as possible
• Includes the quality of transceivers, amplifiers,
  dispersion compensators, filters, etc.
• Ability to easily add/change services as well as
  entire new nodes and fibers
• No disruption to existing users hot swappable,
  optional redundancy and optical protection
• Plug-and-play as simple as popping in
  additional SFPs/XFPs, and connecting up new
  access fibers
• Granularity of single wavelengths for add/drop
• Switching
• Switch the paths of short-/medium-term research
  applications via central management workstation
• Switching done in seconds or minutes, not weeks
• Protection switching, when used, in lt 50 ms
• Option to use an electronic ROADM and/or optical
  ROADM
• Combine the flexibility of optical switching with
  the practical advantage and simplicity of
  electronic transport (performance monitoring,
  loopbacks, etc.)
• Electronic ROADM to enable simpler
  segment-by-segment engineering, avoid complex
  ring engineering
• Simplicity and elegance of mesh networks

11
Cost-effective, Reliable, Multi-ServiceMetro/Regi
onal High-Speed Transport
Services 10GigE, GigE, 10/100 Emerging
40G/100G Fibre Channel ESCON FICON STM-n/OC-n,
E-n/DS-n Video Any protocol
C/DWDM
C/DWDM
Up to 32 wavelengths320 Gbps capacity (32 x
10G)Up to 600 km
Up to 8 wavelengths (8 or 16 GigE/1G FC) 40 Gbps
capacityUp to 120 km
8600 NMS
C/DWDM
C/DWDM

• 30-40 capital operations savings on
  end-to-end solutions
• Transparent bit-rate/protocol independent
  transport
• 10 GigE LAN PHY transported natively
• Carrier-class reliability
• Comprehensive, open network/element management
• Easy to install, engineer, manage

Carrier-class products transport products at
enterprise prices!
12
Meritons Vision of theEnd-to-end Transport
Network
Metro
CWDM
Access
DWDM
CWDM
DWDM
DWDM
Regional
Metro
DWDM
Metro
CWDM
DWDM
CWDM
Access
Metro

• Mix CWDM and DWDM segment-by-segment. Easier
  segment-by-segment ring engineering.
• CWDM segments up to 120 km unamplified at GigE
  (80 km at 2.5G).
• DWDM reach of 600 km miles with no re-gen only
  amps and DCM required.
• Raman amps for longer reach
• Mix of 10G, 2.5G, 1G wavelengths on the same
  fibre.
• Sophisticated, integrated, managed amps
  dispersion compensation.
• Comprehensive, central/remote network and element
  management.

A cost-effective, switched, multi-service,
transparent wavelength network end-to-end from
access to metro to regional.
13
Meritons Vision of theEnd-to-end Transport
Network
Metro
Access
Metro
Regional
Metro
Access
8600 NMS
Metro

• Mix CWDM and DWDM segment-by-segment. Easier
  segment-by-segment ring engineering.
• CWDM segments up to 120 km unamplified at GigE
  (80 km at 2.5G).
• DWDM reach of 600 km miles with no re-gen only
  amps and DCM required.
• Raman amps for longer reach
• Mix of 10G, 2.5G, 1G wavelengths on the same
  fibre.
• Sophisticated, integrated, managed amps
  dispersion compensation.
• Comprehensive, central/remote network and element
  management.

Any topology, including fully meshed networks,
or hybrid ring/mesh networks, etc.
14
Introducing the OADX
Efficiency and Transparency of an OADM
Leading edge support for metro/regional
high-speed services.
15
The Value PropositionScalability and Cost
Savings
Meritons OADX Solution
Incumbent Vendor Solution
70 Km
70 Km
25 Km
25 Km
40 Km
40 Km
40 Km
40 Km
Scalability delivering up to 70 CAPEX Savings
16
Meriton Metro/RegionalProduct Family

• 3300 OSU
• 40 Gbps capacity
• Any input MM 850 nm, SM 1310 nm or 1550 nm
• CWDM DWDM
• Carrier-class redundancy
• 6 RU (10.5)

• 7200 OADX
• 320 Gbps capacity
• Any input MM 850 nm, SM 1310 nm or 1550 nm
• CWDM DWDM
• Carrier-class redundancy
• 21 RU (36.75")

17
Pluggable TransceiversSFPs and XFPs

• Standardized Multi Source Agreement Packaging
• SFPs 100 M to 2.7 Gbps support
• Any protocol
• XFPs 10G
• 10GigE LAN PHY, 10GigE WAN PHY
• STM-64, OC-192
• Change speed/protocol in software
• Types
• Single wavelength
• 850 nm MM 500 m reach
• 1310 nm SM up to 40 km reach
• 1550 nm SM up to 80 km reach
• CWDM SM 40, 80, and 120 km reach
• DWDM SM 40, 80 km reach

5.5 cm x 1.5 cm x 0.9 cm
7.6 cm x 1.8 cm x 0.8 cm
18
Choosing Between CWDM and DWDM
C-Band
L-Band
CWDM Course Wavelength Division Multiplexing

• 20 nm wavelength spacing
• 8 Channels over Single Mode Fibre (SMF)

DWDM Dense Wavelength Division Multiplexing

• 0.8 nm wavelength spacing
• Also referred to as 100 GHz spacing
• Some products also have 200 GHz spacing half as
  many wavelengths in the C-band (i.e. 16)
• Some long-haul system have 50 GHz spacing twice
  as many waves in the C-band (i.e. 64)
• 32 Channels over SMF (100 GHz)
• 1 Channel of OSC

32 Channels OSC
19
AmplificationCWDM vs. DWDM
80 km
80 km
Requires 1 amplifier per wavelength
CWDM wavelengths

1 EDFA amplifies all wavelengths in the C-band
EDFA
C-band
(DWDM wavelengths)

Requires 1 amplifier per wavelength
L-band

• EDFA Erbium-doped Fibre Amplifier
• DWDM is typically used for longer distance
  transport, because EDFA amplifiers enable very
  long spans more cost-effectively than CWDM.
• Amplifiers typically cost approximately US 20k
  or more

20
Electronic ROADM

• Native signal transparency
• Bit rate and protocol independent
• Fully non-blocking wavelength switching
• Single wavelength granularity
• No stranded wavelengths
• Electrical OEO approach allows for important
  system/network functionality
• Multi-degree support
• Any-to-any grid interconnect (e.g. C to DWDM)
• Wavelength conversion for all channels
• 3R at every node (i.e. Engineers like SONET/SDH)
• Layer 1 Performance Monitoring (PM)
• Multicast lightpaths

7200 OADX
21
Wavelength Switching Cost Sweet Spots
NoteFor 2-degree metro ring applications.
ChannelRate
Optical ROADM
ElectricalOEO
10G
Optical ROADM
ElectricalOEO
2.5G
4
8
12
16
20
24
28
32
Pass-through Channels
22
Optical ROADM Wave-blocker
Wave-blocker
Splitter
Coupler
Drop Filter
Add Filter

• Drop and Add Filters must be tuneable for maximum
  flexibility.
• Hitless filter tuning is a problem.
• Many discrete components so expensive
• High insertion loss Limits DCM Limits reach
  between nodes for fully transparent networks.

23
Optical ROADM Wavelength Selective Switch (WSS)
WavelengthSelective Switch
Coupler
OptionalExpansionPort
DropChannels
Add

• Fewer discrete optical components
• Fully flexible colourless add/drop
• Lower insertion loss
• Limited number of drop ports Use expansion port
  !

24
How Much Capacity ?
100Gbps Duo-binary Wave-locker 1b/s/Hz 16 symbol levels 4 bits per symbol required. 256 symbol levels 8 bits per symbol required.
40Gbps NRZ/CS-RZ/ Wave-locker 10G overlay ? 0.4b/s/Hz Duobinary Wave-locker 0.8b/s/Hz 16 symbol levels 4 bits per symbol
10Gbps No issue NRZ 0.1b/s/Hz Reduced reach Wave-locker NRZ 0.2b/s/Hz Reduced reach No ROADMs Wave-locker 0.4b/s/Hz
100GHz 50GHz 25GHz
25
Optical Link Engineering Methodology
Meriton Uses 2 Software Tools to Design Optical
Amplifier Links

• Allows fast network design and link performance
• calculation
• Customized for Meriton 7200 OADX link endpoints
• and the Meriton 1450 and 1650 family of OFAs
• Estimates Q, OSNR, and Margin
• Can model 2.5G or 10G datarate per wavelength
• Can model of wavelengths per link
• Assumes fixed Impact of non-linear network
  effects
• for all DWDM wavelengths

• Used to determine the actual level and impact of
  
• non-linear effects on the proposed Meriton OFA
  Link Design
• Offers more detailed graphical results of DWDM
• link performance

26
Optical Link Engineering Tools
27
Technologies for Dynamic Optical Networks

• GMPLS standards are still evolving for optical
  networks
• Growing interest for dynamic lightpath
  configurations
• Meritons path management includes a number of
  GMPLS concepts
• OSPF routing on NEs (used for management network
  today)
• GMPLS LMP for auto network discovery, lightpath
  testing, and cable mis-wiring
• Meriton will implement GMPLS in step with
  customers key requirements for mesh networking
• Pre-provisioned shared protection (enabled by
  GMPLS signaling)
• Dynamic (best-effort) signaled protection
• Operator signaled lightpaths (S-LPs)
• Client on-demand wavelengths (O-UNI signaling)
• Participation in initiatives such as Internet2
  HOPI, CANARIE UCLP, etc., is critical

28
Best in Class Network Management

• Automatic Discovery
• Automatic node topology discovery
• Automatic card detection
• Automatic fiber connectivity discovery
• Automatic detection of fiber miscabling
• Powerful Lightpath Provisioning
• Both Operator-Selected Routing or Automatic
  Lightpath Routing
• End-to-end lightpath protection or protection
  only for segments of lightpath
• Non-disruptive Live Lightpath Routing Changes
• Fast Identification and Guided Resolution of
  Fiber Miscabling

The considerable investment Meriton Networks has
made in network management is evident! Managing
Optical Networks Report
29
8600 NMS User Interface
30
8300 EMS GUI
Element cross-connect status
Element status
No navigation frame. Single element only
Cross-connect highlighting
Per element alarm view
31
IVFN Intelligent Virtual Fiber Networks
32
Thank Youbrian.pratt_at_meriton.comgerard.owens_at_mer
iton.com