Joint ITU-T/IEEE Workshop
on The Future of Ethernet Transport
(Geneva, 28 May 2010)
Carrier Perspectives on Moving
Beyond 100G
Martin Carroll
Network & Technology
Verizon
Geneva, 28 May 2010
Putting it into Context
http://xxx.yyy.zzz
IP End-to-End
End-to-End
End
to End
Reliable Connectivity
Ethernet or MPLS-TP
SONET
OTN
ROADM
• 50M wireline customers in 24 states
Serving
Residential
Enterprise
Wireless
•
•
•
•
•
10M residential broadband; 3M FiOS TV customers (available to 11.7m)
92M wireless
i l
customers;
t
llargestt 3G network
t
k
672K route miles of fiber globally; first commercial 100G deployment
200 data centers in 22 countries; IP services in 2700 cities / 159 countries
Serve 98% of the Fortune 1000 customers; #1 provider to U.S. Government
Target Architecture
Control Plane
E-Line
E
Line
Services
EVDO
LTE
LH-OTP
Router
Interconnect
P-OTP
T1
DS3
OC-3
OC-12
1 GbE
10
GbE
Local Node
P-OTP
P-OTP
LH-OTP
P-OTP
ADM
P-OTP
BEAS
T1
10/100
EN
GbE
Video
Local Node
OC-48/
OC-192
P-OTP
Ethernet
Services
P-OTP
LH-OTP
WDM
xPON
O
OLT
BW
Services
LH-OTP
P-OTP: Packet Optical Transport Platform (ROADM + SONET + Ethernet)
LH-OTP: Long Haul Optical Transport Platform (ROADM + OTN + MPLS-TP)
MPLS-TP
Switching
LH-OTP
LH-OTP
λ
Services
Transported Services
Time Division Multiplexing (TDM)
g
p
Integrated
Optical
Service
Packet
Services
Ethernet
100G
10G
1G
Converges
Packet, TDM, Wavelength
Technologies
100M
Wavelength
Services
ODU-1, ODU-2, ODU-3, ODU-4
OC
OC-n
TDM Services
10Mbps
100Mbps
1,000Mbps
Storage
DS-1
DS-3
10M
Fast Ethernet
50Mbps, 100Mbps
Gigabit Ethernet
50Mbps, 100Mbps, 150Mbps,
300Mbps, 450Mbps, 600Mbps &
1000Mb
1000Mbps
10/40 Gigabit Ethernet
Ethernet Packet Ring Service
10M
1M
DS-1/DS-3
OC-3/3c
OC-12/12c
OC-48/48c
OC-192/192c
1Gbps & 2Gbps Fiber Channel
Digital Video
270Mbps SDI
1.485Gbps HD-SDI
100G Field Trials
Video signal displayed at Miami
Relative Intens
sity (dB)
0
Video signal fed to
LambdaXtreme® at
Tampa
-10
10G
100G
-20
-30
-40
186.5
187
187.5
188
188.5
189
Frequency (THz)
Nortel OME 6500 with 100G cards
Trial equipment in Lab 400, Richardson, TX
PUMPA
PUMPB
PUMPC
MCU
MIB32
OSCTUI
PUMPC
PUMPB
PUMPA
MCU
MIB32
PUMPA
PUMPA
PUMPB
PUMPC
OSCTUI
OLIINC
OLIINC
10G
test set
Cable compartment
2xUDCM in tray
Bay
100G Rx (PM-QPSK)
hiT 7500 system
10G
40G
40G
10G
X
Laser
Y
π/2
π/2
Marsh
all, TX
- X0
50G
- X90ADC/
data Signal
- Y90
storageprocessing
- Y0
(off-line)
50 GH 50-GHz
50-GHz
50
M
Mux
GH D
Demux
QP
90o
SK
100G TxX
100G Rx
hyb
QP
Fil X
Laser 1 mo
rido
SK
90
d.
ter
FE
Y hyb
Data Y mo
C
Laser 1 rid
proce d.
QP
Laser 2 90o
ssing codi
hyb
SK
ng
XQ
Fil X
QP
rido
mo
90
Laser 2 SK
ter
d.
Y hyb
Y mo
rid
d. transmitter 40G (CP-QPSK) receiver
40G (CP-QPSK)
10G (OOK) transmitter
10G (OOK) receiver
o/e
o/e
A/ o/e
A/
A/ o/e
A/
DDS
DDSPD
SPD
FEC decoding and
da
ata re-alignment
Bay
PUMPC
PUMPB
PUMPC
PUMPB
PUMPA
MTS SR1 AF Shelf
PUMPA
PUMPB
PUMPC
2xUDCM in tray
MTS SR1 AF Shelf
MCU
MIB32
Cable compartment
2xUDCM in tray
OSCTUI
Cable compartment
Cable compartment
2xUDCM in tray
100G Tx (PM-QPSK)
10G
10G
10G
10G
40G
Data
Laser
generator
40G
10G
10G
10G
10G
OLIINC
OLIINC
PUMPA
PUMPB
PUMPC
PUMPC
PUMPB
PUMPA
MCU
MIB32
OSCTUI
OLIINC
OLIINC
PDP
MTS SR1 AF Shelf
OLIINC
OLIINC
Dallas Field
Fiber Loop
PDP
MTS SR1 AF Shelf
Longvi
ew,
TX
Verizon Business long-haul site at Longview, TX
Multi-Vendor 100G Trial
10GE
10GE
Tester
Router 1
100GE card CFP
100GE card CFP
Router
2
100G
Transponder
100GE
Tx
Rx
Tx
Tx
Rx
CFP
Rx
10GE card
GE
Video
Decoder
Video
Display
Video
Encoder
Video
signal
112 Gb/s
ROADM
80-km
Dallas
g
Ring
80-km
Dallas
Ring
x9
Router:
Client I/F:
Transport:
Physical media:
Traffic verification:
Juniper T1600 router with 100GE cards
Finisar LR4 industry standard 100G CFPs
NEC 100G transponders and ULH DWDM system
1520Km of field deployed fiber in DFW metro area
10GE test set and High Def encoded video test traffic
100G Deployment
100Gb/s Live
Frankfurt
Frankfurt
Paris to Frankfurt
893km route
13 spans
No Regeneration
Operational
December 2009
Adding
g 100Gb/s
/
routes in 2010
Hartwald
Alsbach
Karlsruhe
Buhl
Strasbourg
WSS ROADM
Keskastel
Line Amplifier
Metz
(Glass Thru)
Skipped (Glass-Thru)
Fresnes
en Woevre
Raman Link
Saint Menehould
La Veuve
Paris
Soucy
Sommesous
Troyes
Montereau
100Gb/s Roll-out Underway !!
Thirst for More
Unrelenting network traffic growth pressure
bandwidth increases
Higher speed data services; symmetrical
Commercial/residential video, HD/3D-TV, etc.
Fixed/mobile convergence; service mobility
Distributed computing; SANS
Route/service diversity
Wireless backhaul~ 3G, 4G explosion
Demand for bundled services and access
upgrades will trigger global FTTH/B subscriptions
to reach 183.9 million by 2015 [Global Industry
Analysts, Inc.]
Forecast 10G/40G/100G port shipments together
increasing 10-fold from 2009 to 2014 [Infonetics]
Carriers will look to higher speed options
Considerations
Higher rate needed in 2015/2016 timeframe to
meet traffic projections
p j
Large customer/carrier’s carrier services place
additional requirements
Major
j network changes
g will be required
q
to
support rates over 100G
Incompatible with existing ROADMs
Maximize new rate for demand window
Line side vs. client side
Same or different rate step
Distance requirements may not be the same
Weigh implications, cost, availability timeframes
Line Side
Client Side
Bandwidth Scalability
Amplifier
Bandwidth
Increasing spectral efficiency
is the logical first step
because it can be applied to
existing line systems
O,E,S,C,L
, , , ,
Bands
C+L
Bands
Spectral
S
t l
Efficiency
C Band
100 GHz
50 GHz
25 GHz
Channell
Ch
Spacing
NRZ DPSK
PM
QPSK
PM
M-QAM
Adding amplifier bandwidth
seems reasonable for next
generation systems assuming
Raman amplification is
needed anyway
Reducing channel spacing
appears to be the best long
term avenue for scalability
assuming coherent
h
detection
d
systems
40 Gb/s λ
40 Gb/s λ
1568.77
1568.36
1567.95
40 Gb/s λ
10 Gb/s λ
1569.18
40 Gb/s λ
40 Gb/s λ
1569.59
…
100 Gb/s
sλ
10 Gb/s λ
Flexible Bandwidth Grid
1531.12
1530.72
1530.33
Current generation
WDM systems
supported 10G, 40G
and now 100G in
service upgrades
λ, nm
50GHz
75GHz
75GHz
100 Gb/s λ
400 Gb/s λ
400 Gb/s λ
1569.59
1569.18
1568.77
1568.36
1567.95
150GH
150GHz
1 Tb
b/s λ
50GHz
100 Gb/s λ
C‐band, fixed 50 GHz grid
…
1531.12
EExample
l off a future
f t
C‐band,
C b d 50‐200
50 200 GHz
GH
Flexible Grid (in 25 GHz increments)
1530.72
1530.33
λ, nm
Future systems will be
expected to support in
service upgrades to at
least 1T / channel
Given the current view,
more bandwidth is the
only way to achieve
this w/o paying a
significant reach
penalty due to OSNR
requirements
Flexible Layer 1 Network
A
B
150 GHz
100G
1T
Enables improved applications
Network Defragmentation
B d idth Adjustments
Bandwidth
Adj t
t
Network Maintenance
Network Restoration
C
D
E
100G
1T
T
50 GHz
Colorless/Directionless/Contentionless ROADM node with flexible grid
Colorless wavelength add/drop with directional routing
Choose the bandwidth of the light path to match the service bitrate
Use multiple copies of the same color wavelength on the add/drop structure
What’s Next?
C-band only for comparis
son
Total capacity-d
T
distance (Pb/s-k
km)
1000
Optimistic
100
?
10
1
Source: [1-10]
Pessimistic
0.1
10
100
1000
10000
Data rate per channel (Gb/s)
• 400Gb/s may not be the right answer…
Homework for
Industry
Academia
Government
•
•
•
•
•
Is 1Tb/s per channel an achievable target?
If so, what would be the best modulation format?
Can DSP based coherent receivers continue to scale?
Would revisiting the ITU-T DWDM grid solve the problem?
Is spectral efficiency worth the complexity of variable width channels?
Summary
100G is a critical technology; beyond 100G will be required for
transport scalability
Flexible bandwidth grid architectures are required for
upgradability
Colorless, directionless & contentionless are required for flexibility
TDM type connectivity is required for PL and EPL serviceability
MPLS-TP is required
d for
f optimized
d router connectivity
Integration is required to improve cost and reliability
At the end of the day
day, we need a cost effective solution that gives
us both scalability and backwards compatibility to some extent
Due diligence is needed ~
Think through all the implications before plotting a path to get there
We should not make decisions on what comes next until there is
experience to be bored with 100G – or we may sell ourselves short
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10
10.
D. G. Foursa et al, "2Tb/s (200x10Gb/s) Data Transmission over 7,300km Using
150km Spaced Repeaters Enabled by ROPA Technology," OFC/NFOEC 2007, PDP25.
C. Laperle et al, "Wavelength Division Multiplexing (WDM) and Polarization Mode
Dispersion (PMD) Performance of a Coherent 40Gbit/s Dual-Polarization Quadrature
Ph
Phase
Shift
Shif Keying
K i
(DP-QPSK)
(DP QPSK) T
Transceiver,"
i
" OFC/NFOEC 2007
2007, PDP16.
PDP16
M. Salsi et al, "155x100Gbit/s coherent PDM-QPSK transmission over 7,200kmALU
100G," ECOC 2009, PD2.5.
H. Masuda et al, "13.5-Tb/s (135 x 111-Gb/s/ch) No-Guard-Interval Coherent
OFDM Transmission over 6,248 km Using SNR Maximized Second
Second-Order
Order DRA in the
Extended L-Band," OFC/NFOEC 2009, PDPB5.
S. L. Jansen et al, "10x121.9-Gb/s PDM-OFDM Transmission with 2-b/s/Hz Spectral
Efficiency over 1,000 km of SSMF," OFC/NFOEC 2008, PDP2.
J. P. Turkiewicz et al, "Field trial of 160 Gbit/s OTDM add/drop node in a link 275
km deployed fiber,"
fiber " OFC 2004,
2004 PDP1.
PDP1
P. Guan et al, "Low-Penalty 5 320 Gb/s/Single-Channel WDM DPSK Transmission
Over 525 km Using Time-Domain Optical Fourier Transformation," IEEE PTL,
21(21), 1579 (2009).
A.I. Siahlo et al, "640 Gb/s OTDM Transmission and Demultiplexing using a NOLM
with Commercially Available Highly Non-linear Fiber," CLEO 2005, CTuO1.
Y. Ma et al, "1-Tb/s single-channel coherent optical OFDM transmission over 600km SSMF fiber with subwavelength bandwidth access," Opt. Express 17, 9421-9427
(2009).
M Nakazawa et al,
M.
al "1
1.28
28 Tbit/s-70 km OTDM transmission using third- and fourthorder simultaneous dispersion compensation with a phase modulator," Electronics
Letters, 36(24), 2027-2029 (2000).