Optical Trends in Premises Networks

1
Optical Trends in Premises Networks

• Russell Ellis, PhD.
• Corning Optical Fiber
• Premises Fibers
• March 7, 2006

2
Agenda

• Traffic demand continues to grow
• Market trend from LEDs to lasers
• Trends in lasers
• The need for advanced fibers and bandwidth
  measurement techniques
• OM3, OM3
• Copper still dominates
• Trends in connectivity
• Whats next in fiber development?
• (40G/100G deployments)

3
Bandwidth Demand is GrowingGlobal Internet
Traffic Expected to Grow
Financial Reporting Sarbanes-Oxley
increasing data management as more sophisticated
tools come to market Security and Regulation
Disaster recovery requirements (primary,
secondary, tertiary) On-line business activity
Internally, connecting remote/global locations.
Externally, connecting customers with suppliers
to drive supply chain efficiency. ERP, CRM, and
other multi-site software development Others
Digital media and document/medical imagery, high
cost of downtime
Source RHK Inc (Jan 2005)
4
Deployment speeds continue to grow 1G becoming
the standard

• 1G almost at price parity with 100M, especially
  in copper
• Move to optical 1G requires market shift from
  LEDs to Lasers
• Growth in 1G links drives 10G interest,
  awareness, and initial backbone deployments

Source DellOro
5
Fiber winning everywhere except horizontal
Single-mode fiber taking share

• Horizontal
• 10/100/1000 Mb/s

• Riser
• 80 fiber and increasing
• 25 1Gb/s - 75 100 Mb/s
• Fiber has won, MMF dominates

• Interbuilding
• 95 fiber and increasing
• 10 Gb/s initial deployments
• 50 1G/50 100M
• Fiber has won, single-mode fiber continues to gain

• Data Centers
• 50 fiber and increasing
• 1, 2, 4 and 10 Gb/s
• Fiber is winning, MMF is winning

Souce Corning Optical Fiber/Corning Cable
Systems Analysis
6
In Optical Premises 850 nm VCSELs dominate at
1G, harder challenges at 10G
90s
00s
80s
850 nm VCSELs have won at 1G cost (rapid scaled
production)
850 nm LEDs attractive for 10 Mb/s, but high loss
What window for 10 Gb/s?
850 nm
1300 LEDs won at 100 Mb/s more manufacturers
supported, better performance
1300 nm

• 1G, then-current generation of fibers able to
  support 10G, required new fiber
  attributes (specifications and measurements)
• OM3 fiber developed for 10G 850 nm some
  competing Ethernet technologies for installed
  base focus on 1300 nm

7
10G was the first time fiber and application
standards were co-developed
Standards Roadmap
Below 10 Gb/s, application standards used
then-current multimode fibers to design network
solutions
However, 10Gb/s required new fiber specifications
and measurement methods
8G
10G
10000
FC
10 GbE Sonet/SDH InfiniBand
4G
2G
FC
1G
1000
FC
Data rate (Gb/s)
GbE Fibre Channel
We expect next step in data rates to use
primarily current fiber specifications
100M
100
Fast Ethernet ATM
10M
10
Lasers
LEDs
Ethernet Token Ring FDDI
1
1985
1999
1995
2002
2006
8
Legacy bandwidth measurements cannot predict
laser performance
Source TIA FO-2.2.1 Round Robin
9
Move from LEDs to lasers required new bandwidth
measurement systems
Light Sources
Bandwidth Measurement

• OFL (Overfilled-Launch)
• Designed to predict performance of low-speed
  LEDs, not lasers
• Power distributed in 100 of the fiber core, like
  LEDs
• Perturbations in index profile undetected
• Laser-Based Measurements
• RML (Restricted Mode Launch) or DMD
  (Differential Mode Delay)
• Power distributed in a narrow region
• Simulates an actual laser launch
• More accurate indication of performance in
  high-speed laser-based systems

(Typically 10 and 100 Mb/s)
(1, 2, 4, 8, 10 Gb/s and higher)
10
Dramatic increase in capability of MMF 10G
fibers growing the fastest
OM3 fiber has grown at over 100 CAGR since 2002
standard
10G Laser-Optimized
1G
1G
Source Corning analysis
11
50 µm will continue to grow 10G the fastest

• Increased speeds will drive higher-performance
  products
• Assumes no impact from EDC

12
Active component costs dominate premises link
economics

• Typical 300 meter backbone

Switch Electronics
10G Tx/Rx (XFP)
Source www.foundry.com, www.peppm.org, Corning
analysis
13
SX (850 nm) lower-cost than LX (1300 nm)

• Assumptions
• 300 m, 24F count cable, 24F Passive Interconnect
  (x2), 18x 1 Gb/s Transceivers
• Key findings
• Cable is a very small portion of link costs
• SX (MMF) solutions always cheaper
• OM3 fiber
• Supports 10Gb/s over 300m
• Lowest cost upgrade path to 10Gb/s

Relative System Costs 1 Gigabit over 300m
2.5
Fiber Cable
Hardware
2
Small ? for 10G capability
1.5
Single-mode
1300nm
1300nm
850nm
1
850nm
850nm
0.5
0
OM-2
OM-3
OM-2
OS1 (LX)
OM-3
SX solution saves 50 over LX
14
What is an OM3 fiber?Why consider installing it?

• MMF, 50 ?m
• Higher Bandwidth
• (i.e. 4700 MHz.km EMB)

• Faster
• At a given length
• 40G or 100G upgrade capability

• Safer
• More bandwidth enables spare margin
• ? 2.8 dBm for
• 10G 300m link
• Providing an additional
• ? 5 connector pairs, or
• ? 50 fusion splices

• Farther
• At a given bit rate
• 550m at 10 Gb/s

15
Device cost typically determines application
segment
10G Serial Alternatives
Relative Laser Cost
Price Sensitive Market?
DFB
Long Haul

Metro
FP

Proposed 10GBASE-LRM
Access
1300 VCSEL

LANs
850 VCSEL
Premises

Low cost 850 VCSELs dominate the price sensitive
Premises market
16
Comparison Spectral Characteristics
LED
60 nm

l
4 nm
Fabry-Perot Laser

Higher Performance
3-10 dB
VCSEL

0.1 nm
DFB laser
30dB

17
We expect lower-cost solutions in premises
networks will be at 850 nm

• 850 nm VCSELs have won the 1G premises market
• 850 nm VCSELs just entering high-volume
  manufacturing cycle
• Will continue to be low-cost solution for 10G
• However, LR (1300 nm) solutions will capture some
  market share in Premises

10G Transceivers
18
LRM/EDC capabilities are uncertain to 220 m

• Stated objective of the LRM standards effort is
  to enable the installed OM1 and OM2 base at 10
  Gb/s
• We are working with other fiber manufacturers to
  accurately characterize the installed base of
  fiber
• Given a fixed functional objective, the EDC
  solution cannot serve 99 of known links while
  also being low-cost

• Low Cost/
• complexity
• PIE-D4.5

Current performance target
These goals are not independent

• Worst Case design
• 99 coverage of installed base

• Functional Objectives
• BER 10-12
• Length220 m

19
Copper continues to dominate premises
Single-mode fiber gaining optical share

• Challenges for MMF
• Copper/Horizontal
• Electronics Cost
• Powering
• Interfaces
• Single-mode fiber
• Futureproof
• Simplicity
• Electronics mfgs influence

18
19
19
17
62
64
Source Corning, CRU
20
Why Fiber?

• Higher data rates and longer link lengths
• Flexible, reliable networks with low latency
• Unparalleled network security
• Immune to EMI, RFI and cross-talk
• Small lightweight cables maximizes pathway and
  space utilization
• Higher port density
• Easier installation, handling and termination
• Simplified field testing
• Longer cable life cycle
• Lower power consumption, less expensive to operate

21
Why Fiber?
22
Fiber connectivity trends

• Pre-terminated solutions 75 faster
  installations than traditional cabling solutions
• Modular component design facilitates moves,
  adds and changes
• Small-form-factor products (LC, MT-RJ, MTP,
  ribbon cable) reduce space needed under floor,
  overhead and in racks/cabinets

• Small diameter cable
• 6 connectors
• 72 terminations

23
Whats next? 40G or 100G

• 40G or 100G
• Not going to be a serial solution
• WDM OM3 will provide extra BW for mux/demux
• Parallel Optics
• IEEE 802.3 call for interest in 2006