100G Technology Presentation

100G Technology
A basic introduction
Dr. Paul Brooks
August 2013
Aim
This presentation aims to help you understand

the following:
The 100G client side

The drivers for 2nd generation 100G
CFP2
New technology challenges
Developing and validating 2nd generation 100G
2012 JDS Uniphase Corporation
JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION
100G flavours
40GE & 100GE

Native Ethernet running at 41.25 Gbs or 103.125Gbs typically
over an LR4 (4 x 10G/25G) optical interface (4 on a SMF),
coding via PCS layer (VL & XLAUI/CAUI)
Line side 100G

Complex modulation => DP-QPSK (dual polarization,
quadrature shift keying)
Coherent detection complex DSP based receivers BUT can
compensate for PMD & CD electronically
100G carried on one coloured lambda, compatible with existing
infrastructure (ROADM, EFDAs etc)
100G Where in the network?

Router
CFP optics
4 x 25G
Router Router via 100GE
Router
Router switch via 40GE
CFP optics
4 x 25G
Likely 4x10G CFP first.

Also CXP/QSFP
Switch
CFP optics 4
x 28G
Router DWDM
Inter-domain OTU4 (or 100GE)
High density,
short reach
40G & 100G
100G Line side

Modulation &
Strong FEC
Legacy interfaces
SONET, 10G
40G serial etc
Cloud computing etc
High density,
10G & 40G
DWDM equipment
Segmentation
Enterprise customers
High port density (especially on faceplate)
Today => 48 x SFP+ = 480Gb/s
Very cost sensitive

Cost of pluggables can dominate
Mostly short reach applications

~90% SR (MM, 850nm, ~100m reach) pluggables
Telecoms
Port density not critical
1:1 mapping client line side
Deeper feature set (Optical XC, OTN, router)

mostly long reach (10 km) pluggables
Higher cost per port
100G Optics
Main current
approach
SR10 MMF
CXP
100GE
Data Centre
Enterprise
100G
LR4 SMF
CFP
Line Side
OIF MSA
10 (SMF)
CFP
100GE
& OTU4
OIF 5 x 7
module
850 nm
Santur (now
Neophotonics)
1550 nm
Next step
CFP2
2012 Beta
CFP4
2014
Lineside
CFP
ASIC inside
CFP2 to replace CFP over time

- Same photonic interface (LR4)
- Electrical bus moving from 10G to 25G
CFP4 high density & lower power

40G Optics
40G
SR4 (100m MMF)

QSFP+
FR (2km serial SMF)

& G.693 CFP
40GE
Data Centre
Enterprise
850 nm
40GE
& OTU3
& OC-768/
STM-256
2011/2012
LR4 (10km SMF)

& G.695 CFP
Legacy (client & line)

300-pin MSA
40GE
& OTU3
& OC-768/
STM-256
QSFP+
2011
Line Side
OIF MSA
OC-768/
STM-64/
OTU3/
OTU3e1/3e2
DWDM
2011+
1310 nm
1550 nm
1550 nm
SMF Support current on CFP; moving soon to QSFP+

Serial CFP may replace 300-pin MSA in multi-rate 40G applications
For use on current fiber infrastructure
Considerations for line side testing with OIF MSA (same FF as 100G)
QSFP+ will dominate new 40G products for LR & SR
Focus Today CFP2
100G SMF Evolution
First Gen CFP is bulky, expensive and power hungry based on 10G electric
interface
Next Gen CFP2/4 is fully based on 25G technology
smaller for higher density to support Terabit line card

- power consumption comparing to CFP
Scalability from CFP (gearbox, MDIO)
lower cost overall
Challenge is integration &

power density
SMF Space
Challenge is 25G I/O
100G LR4
CFP2
100G LR4/10
CFP
2011
2012
2013
100G LR4
3rd Gen
2014
2015
100G Form Factors Driven by Density
Current state of the art

400 Gb = ~56W
~ 2013
800 Gb = ~64W
~ 2014
1.6 Tb = ~80W
~ 2015
3.2Tb = ~160W
But power density & cooling will remain an issue!
Power is for pluggable optics only, total power will be considerable higher
10
CFP Detailed Block Diagram Complexity Revealed!
Photonic integration with

allowed higher density/lower
power TOSA/ROSA but what
about EML vs. DML?
Gearbox moving to CMOS

in 2013
Lower power/cost
Smaller form factor
11
CFP2 simplicity drives down cost & power
Tighter photonic integration

Lower complexity electronics
But challenges in thermal density need to be met for future
12
100G LR4 CFP2 and CFP4

CFP2
~ 9W
CFP4
>6W
Move to 25G I/O electrical interface (direct 1:1 mapping of electrical &
optical lanes).
CFP2 can support native 10 x 10G CFP4 cannot without MLG IC
Gearbox in module replace with CDR
MDIO control S/W re-use from CFP => CFP2 => CFP4
25G electrical I/O is 1st generation technology!
13
CFP2 optical module flavors

Electrical connector
supports 4 x 25G or
10 x 10G operation
Fibre connector will

vary with module
type (MPO/LC)
SR10 vcsel based, lowest cost, lowest

power short reach (300m) MM ribbon
cable
LR4 based on temperature controlled
laser, 10 km reach, price x5 SR10 SM
fibre
SR4 based on 25G vcsel very novel
technology, could offer lower power, low cost
but will require FEC, very short reach MM
ribbon fibre (100m)
LR10 port expander 10 x 10GE ports,
required SM ribbon cable
LR10 (Santur style) 100G carried as 10
lambda on a SM fibre inter-op with
installed base of Santur/Neophotonics LR10
ER4 extended reach (40km) better RX
performance required. Higher power
consumption with price premium
14
CFP2, CFP4 & QSFP28
All of these represent next generation 100G

They all offer increased density (and lower power & cost)
than CFP
All have a part to play in the future 100G ecosystem
CFP => very flexible slot that can support 40G (parallel &
serial), 100G (10x10 & 4x25) and line side applications
CFP2 => native support for 10x10 and 4x25G, good thermal
density
CFP4/QSFP28 => ultra high density ideal for datacentric
applications
CFP2, CFP4 & QSFP28 all require a stable 25G I/O to work
15
MDIO
MDIO is a serial control bus for control and management of

Gigabit Ethernet pluggable optics
It is used in CFP, CFP2 & CFP4
It is laid out in IEEE 802.3 but is described in depth for the
CFPs in the CFP MSA
It is an integral part of any CFP based system and needs to
be tested and debugged.
QSFP28 will use the I^2C bus
I^2C is a lighter bus not as fast or complex as MDIO
CFP, CFP2, CFP4 => MDIO
SFP, XFP, QSFP+, QSFP28 => I^C
16
100G Line-side Module Evolution

OIF Transponder
5 x 7
75W
Price
CFP (32W)
145mm x 76mm
ASIC inside (~20W)
CFP2 (12W)
106mm x 40mm
ASIC outside
Fits on CFP slot (demanding

power envelope) ~32W
Smaller size, lower power dissipation

17
17
Current Challenges for 100G modules
100GE is evolution & revolution

The physical layer (both electrical & photonic is a revolution).
The Ethernet part is an evolution from 10GE
Price point expectations

CFP module prices need to be driven down to allow wider
deployment
CFP does not meet density expectations
ASICS to replace FPGA (port density & cost)
Transition to production & 2nd generation 100G (CFP2 &

25G I/O)
Physical layer will be major challenge 25G is new!
CFP2/4 allows the density & roadmap to cost

expectations
Debate on post CFP2.what role for QSFP2?
18
Physical Layer at 25/28G
1st generation 100G used mature (3rd generation 10G)

technology
Even this caused significant challenges for early adopters
2nd generation 100G will use 25/28G I/O

Still very novel technology (1st generation)
New ecosystem of ICs, connectors, PCBs all need validation
Users need test solutions which give true insight into the
issues of the physical layer.
But denser ICs need framed traffic (Ethernet, OTN etc)
Physical layer applications MUST be protocol aware as
required.
Crosstalk, loss, jitter, bit-slips etc will all need troubleshooting.
More intelligence needed to troubleshoot
Margins become critical
19
Our Pedigree
October 2009
JDSU releases worlds first 100GE & OTU4 test
set. Unique (patented) features include Lambda
mapping, dynamic skew & MDIO debug.
Product now includes comprehensive OTN
mapping from ODUflex upwards.
December 2011
JDSU releases most compact and
comprehensive 100G field solution, includes
native support for CFP & QSFP+
20
October 2012
JDSU announce worlds 1st CFP2 test solution
CFP2 is here now!

It is the form factor to take 100G to prime time
The 25/28G I/O is extremely demanding
New generation T&M is required to move CFP2 and beyond
to successful deployment
21
Paul Brooks
Questions?

100G Technology Presentation

Uploaded by

Copyright:

Available Formats

100G Technology Presentation

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

100G Technology Presentation

Uploaded by

Copyright:

Available Formats

100G Technology

This presentation aims to help you understand

The 100G client side

2012 JDS Uniphase Corporation

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

40GE & 100GE

Line side 100G

2012 JDS Uniphase Corporation

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

100G Where in the network?

Router Router via 100GE

Router switch via 40GE

Likely 4x10G CFP first.

100G Line side

2012 JDS Uniphase Corporation

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

Very cost sensitive

Mostly short reach applications

Deeper feature set (Optical XC, OTN, router)

2012 JDS Uniphase Corporation

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

CFP2 to replace CFP over time

CFP4 high density & lower power

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

SR4 (100m MMF)

FR (2km serial SMF)

LR4 (10km SMF)

Legacy (client & line)

SMF Support current on CFP; moving soon to QSFP+

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

Focus Today CFP2

2012 JDS Uniphase Corporation

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

100G SMF Evolution

smaller for higher density to support Terabit line card

Challenge is integration &

Challenge is 25G I/O

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

100G Form Factors Driven by Density

Current state of the art

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

CFP Detailed Block Diagram Complexity Revealed!

Photonic integration with

Gearbox moving to CMOS

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

CFP2 simplicity drives down cost & power

Tighter photonic integration

2012 JDS Uniphase Corporation

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

100G LR4 CFP2 and CFP4

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

CFP2 optical module flavors

Fibre connector will

SR10 vcsel based, lowest cost, lowest

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

CFP2, CFP4 & QSFP28

All of these represent next generation 100G

2012 JDS Uniphase Corporation

JDSU CONFIDENTIAL AND PROPRIETARY INFORMATION

MDIO is a serial control bus for control and management of