Escolar Documentos
Profissional Documentos
Cultura Documentos
Technology Overview
Applications
LAN / WAN PHYs; Optics; Layers
Technology Overview
Operations, Administration, & Management (OAM)
Point to point (P2P)
Ethernet over unclassified copper (EDSL; EFMCu)
Point to multi-point (P2MP; EPON)
Technology Overview
Structure
Access
Fairness
Protection
Comparison
Microtrenching
Microconduit
g
g
Towards Simplification
Towards higher speed; lower cost vs. Moores Law
Ethernet to the rescue in the Access Space
QOS and OAM can be and must be solved
Economic models can support True Broadband
Services
Distractions or complements
Federal regulation and policy will be the single
greatest influence on technology development
Investment as a positive feedback system
g
g
g
Is anyone on line?
n
n
Dedicated Media
UTP
CSMA/CD:
UTP
Carrier Sense
Multiple Access
with Collision
UTP
UTP
Detection
Half-Duplex 10BASE-T (Star Topology, UTP cable, 1990)
Bridge/Switch
UTP
UTP
Collision-Free
Source: Luke Maki, Boeing Corporation, 2002
Full Duplex 10/100BASE-T (1992/1993)
Dedicated Media
UTP
UTP
UTP
UTP
UTP
Collision-Free
Source: Luke Maki, Boeing Corporation, 2002
Full Duplex 10/100BASE-T (1992/1993)
Dedicated Media
UTP
UTP
UTP
UTP
UTP
Collision-Free
Full Duplex 10/100BASE-T (1992/1993)
Ethernet:
Application
Presentation
Session
Reconciliation Sublayer
Transport
(XG) MII
Network
Datalink
Physical
802.10
SILS
IEEE
OSI
Layer
2
802
Overview
Architecture
802.1
Management
802.3
802.4
CSMA/CD TBUS
ISO
Ethernet
ISO
802.5
802.6
802.9
802.11
802.12
802.14
802.15
802.16
802.17
TRING
DQDB
ISLAN
WLAN
DPAP
CATV
WPAN
BWA
RPR
OSI
Layer
1
ISO
IEEE 802.7-1989
IEEE 802.8-1987
802.5
802.6
802.7
802.8
802.9
Fair
Weighting
802.1Q
VLAN
Tagging
Bandwidth
Shaping
Packet
Forwarding
Engine
Switching
Table
802.3 CSMA/CD
MAC
PHY
MAC
PHY
MAC
PHY
Buffering
MAC
PHY
MAC
PHY
MAC
STA
Station
MDIO
PHY
MMD
WAN
MAN
EAN
LAN
Preamble
Start of Frame Delimiter
Destination Address
Source Address
Length / Type Field
MAC Client Data
Pad
Frame Check Sequence
Extension
Preamble
Start of Frame Delimiter
Destination Address
Source Address
VLAN Tag (802.3ac)
Length / Type Field
MAC Client Data
Pad
Frame Check Sequence
Extension
Standards Board
RevCom
Review Committee
IEEE 802
Sponsor Group
NesCom
New Stds. Committee
IEEE 802.3
Working Group
IEEE P802.3af
Task Force
IEEE P802.3ah
Task Force
Process in Summary
g
g
g
g
g
g
The 5 Criteria
1. Broad Market Potential
Broad set(s) of applications // Multiple vendors, multiple users
balanced cost, LAN vs.. attached stations
3. Distinct Identity
Substantially different from other 802.3 specs/ solutions
Unique solution for problem (not two alternatives/ problem)
Easy for document reader to select relevant spec
4. Technical Feasibility
Demonstrated feasibility; reports -- working models
Proven technology, reasonable testing // Confidence in reliability
5. Economic Feasibility
Cost factors known, reliable data // Reasonable cost for performance
expected // Total Installation costs considered
HSSG
FORMED
STD!
WG
Ballot
LMSC
Ballot
TF
Review
PAR Approved
N
O
V
J
A
N
M
A
R
M
A
Y
J
U
L
Y
S
E
P
T
Objectives Drafted
N
O
V
J
A
N
M
A
R
M
A
Y
J
U
L
Y
S
E
P
T
PAR Drafted
Last New
Proposal;
Adopt
Core
Proposal
Last
Feature
Last Technical
Change
N
O
V
Draft 5
S
E
P
T
Draft 4
J
U
L
Y
Draft 3
M
A
Y
2001
Draft 1
M
A
R
2000
Draft 2
1999
J
A
N
M
A
R
S
E
P
T
Objectives Drafted
N
O
V
J
A
N
M
A
R
M
A
Y
J
U
L
Y
S
E
P
T
Par Drafted
Last New
Proposal;
Adopt
Core
Proposal
Last
Feature
Last Technical
Change
N
O
V
J
A
N
M
A
R
M
A
Y
Draft 4.1
Draft 4.2
Draft 4.3
Draft 5.0
J
U
L
Y
Draft 3.3
Draft 3.4
Draft 4.0
M
A
Y
Draft 3
M
A
R
Draft 2.1
J
A
N
Draft 2
N
O
V
Draft 1
S
E
P
T
J
U
L
Y
2001
Draft 3.2
2000
Draft 1.1
1999
M
A
Y
LMSC
Ballot
TF
Review
PAR Approved
M
A
R
STD!
WG
Ballot
Draft 3.1
HSSG
FORMED
J
U
L
Y
Approved
Standard
D4.0
D4.1
D4.2
D4.3 D5.0
Return
76%
79%
83%
85%
87%
Abstain
8%
5%
5%
5%
4%
Approve
82%
82%
86%
88%
96%
NOT
YOUR
FATHERS
ETHERNET
Enterprise Networking
g
Dumb terminals
n
g
g
g
g
g
g
attached to
mainframes
Star wired
Relatively short
distances
High reliability
Easy to maintain
Lowest cost (?)
Mission critical
Note: IEEE 802 formed in 1980
circa 80
Ethernet CSMA/CD
g
g
g
g
g
g
g
g
circa 83
Simplex operation
Enterprise Networking
g
g
g
Dumb terminal
emulation cards in PCs
Still mission critical
Enter LOTUS 1-2-3
Sneakernet
circa 85
Enterprise Networking
g
g
g
g
Dumb terminal
emulation cards
still in PCs
(mission critical)
Ethernet cards
also (PC-based SW
becoming mission
critical)
> 2x the work
< the reliability
> 2x the expense
circa 86
Ethernet Hubs
g
g
g
g
g
g
circa 85-86
Star wired
Point-to-point only
No shared media
But, protocol behaves
like shared media
Increased distance
Higher reliability
Easier to maintain
Easy upgrade path
Higher cost
10
10
10
10
10
10
Switched Ethernet
g
Full Duplex
n
n
n
n
n
g
g
g
g
g
g
circa 87
No collisions!
Star wired
Point-to-point only
No shared media
Transmitter does not
monitor Rcvr
Increased distance
Highest reliability
Easiest to maintain
Easiest to upgrade
Higher cost
Higher performance
10
10
10
10
10
10
Repeater Set
DTE
MAU
Repeater Set
Repeater Set
10BASE-T Link
Segments
DTE
MAU
g
g
g
g
FOIRL
10BASE-T
g
g
g
circa 90
Inexpensive media
Inexpensive ports
Installation ease
10
10
10
10
10
10
Enterprise Networking
g
g
circa 90
10
10
10
10
10
10
10
100
10
10
100
100
10
10
10
10
100BASE-FX
26.2 Functional Specifications
g
Characteristics
g
g
g
g
g
1000
Gigabit Ethernet
IEEE 802.3z
g
g
g
IEEE 802.3ab
g
Sorry ATM
See: http://www.10gea.org/Tech-whitepapers.htm
1000BASE-T
IEEE 802.3ab
g
Link Aggregation
IEEE 802.3ad
g
10 Gigabit Ethernet
1 of 2
IEEE 802.3ae
g
10 Gigabit Ethernet
2 of 2
g
g
g
g
g
g
g
Sorry Who?
Overview of DTE
Power
IP phone
Web cam
Wireless access point
Security, lighting, HVAC controls
Enables many new types of devices
1 of 3
No wall warts
No expensive AC power wiring for wireless access points
Project Status
n
n
n
n
2 of 3
3 of 3
IP Phone
The Ethernet shaver!
Overview of
Ethernet in the First
Mile
P2P Focus
Hybrid Fiber/Copper
Next-generation, high-speed architectures
EFM copper for the last 700 to 800 meters
Minimum 10 Mbps higher if possible
High bandwidth for entertainment client/server
For stepwise buildout to work, EFMCu must support next-gen
applications
P2P EFMF 1000 or 100 Mbps
P2MP EPON 32 Mbps per ONU
06/03/2002
Source: EFMA 2002
OAM Operations
g
Link Monitoring
Remote Loop-back
Capability Discovery
10 Gigabit Ethernet
in Detail
Serial
LAN PHY
(64B/66B)
WWDM
LAN PHY
(8B/10B)
WWDM
PMD
1310 nm
-LX4
Serial
PMD
850 nm
-SR
Serial
WAN PHY
(64B/66B + WIS)
Serial
PMD
1310 nm
Serial
PMD
1550 nm
-LR
-ER
Serial
PMD
850 nm
-SW
Serial
PMD
1310 nm
-LW
Serial
PMD
1550 nm
-EW
1 Gigabit Ethernet
g
g
g
g
g
g
g
g
g
g
g
g
Misunderstanding Ethernet
AUGUST 14, 2000
g Running Ethernet over WANs may sound like a
nice idea in principle, but its tough to pull off in
practice. One of the fundamental rules about
Ethernet is that the faster the network runs,
the smaller the network gets.
g At 10 Gbps, you end up with a very small
network indeed extending a couple of hundred
yards over multimode fiber, max.
WRONG
802.3ae
LAN MAN RAN WAN
LAN PHY
WAN PHY
!! ! !
!! ! !
10 100
1,000
10,000
10 Gigabit Ethernet
Gigabit Ethernet
Fast Ethernet
Ethernet
Bandwidth (Mbps )
Bandwidth/Distance Evolution
0.1
1
10
Distance (kilometers)
100
1000
62.5 MMF
50 MMF
SMF
160
200
400
500
2000
28m
35m
69m
86m
300m
10km
40 km
240
300m
10 km
300m
@500MHz*km
10 GbE Applications
10GBASE-LR, -ER
Metro Link
Campus X
10GBASE-LR,
-ER, -LW, or -EW
Metro Link
DWDM Optical
Network
10GBASE-LX4
or -LR
Campus Link
10GBASE-LW or -EW
Metro Link
Campus Y
10GBASE-SW
Jumper
Enterprise B
Enterprise A
10GBASE-SR
Jumper
Server
Farm
Enterprise C
Campus B
Campus A
10 GbE in:
SP data centers
& enterprise LANs
n
n
n
n
Switch-to-switch
Switch-to-server
Data centers
Between buildings
10GbE
10GbE
10GbE
10GbE
Internet
Extranet
Data Center
Server
Farm
Location A
10GbE
Service Providers:
g
DWDM mux
Location C
Remote
Servers
10GbE
10GbE
Location B
10GbE
Metro
Metro
10GbE
Metropolitan
Networks
10GbE
10GbE
Location C
Location B
Remote
Servers
10GbE
10GbE
Service Provider
Point of Presence
(PoP)
10GbE
Carrier
Central
Office (CO)
Carrier
Central
Office (CO)
Optical
Transport
Optical
Transport
Service Provider
Point of Presence
(PoP)
10GbE
Core DWDM
Optical Network
National
Backbone
Layer Model
P802.3ae LAYERS
OSI
Reference
Model
Layers
Higher Layers
LLC
MAC Control
MAC
Application
Presentation
XGMII
Session
XGMII
XGMII
64B/66B PCS
Transport
64B/66B PCS
WIS
8B/10B PCS
Network
PMA
PMA
PMA
Data Link
PMD
PMD
PMD
Physical
MDI
MDI
MDI
MEDIUM
MEDIUM
MEDIUM
10GBASE-R
10GBASE-W
10GBASE-X
Device Nomenclature
!
!
!
!
!
!
10GBASE-SR
10GBASE-SW
10GBASE-LR
10GBASE-LW
10GBASE-ER
10GBASE-EW
10GBASE-LX4
15
5
Se 0nm
ria
l
13
1
Se 0nm
ria
l
13
W 10n
W m
DM
85
Se 0nm
ria
l
IS
Optics
W
64
B/
PC 66B
S
Device
8B
/
PC 10B
S
Logic
!
!
!
!
!
!
!
!
10GBASE-X
MAC
10 Gbps
XGMII
10GBASE-X
PMA
8b
8b
8b
8b
8B/10B
Encoder
8B/10B
Encoder
8B/10B
Encoder
8B/10B
Encoder
10b
10b
10b
10b
12.5 Gbps,
4 @ 3.125 Gbps
SERDES
SERDES
SERDES
SERDES
12.5 Gbps,
4 @ 3.125 Gbps
10 Gbps
10GBASE-R Serial
Ethernet Packet + Min. IPG
MAC
10 Gbps
XGMII or XAUI
10.0 Gbps
64B/66B PCS
64-bit Scrambler
Sync. Bits (2)
64b
10.3 Gbps
XSBI
PMA
SERDES
10.3 Gbps
10X
g Further:
40 km (expect proprietary
extensions or WAN)
g Format:
g Management:
Consistent
10GBASE-W Serial
Ethernet Packet + Min. IPG
MAC
Extra IPG
10 Gbps
XGMII or XAUI
64B/66B PCS
9.29 Gbps
Extra IPG Dumped
64-bit Scrambler
Sync. Bits (2)
WIS
64b
9.58 Gbps
9.95 Gbps
XSBI
PMA
SERDES
9.95 Gbps
Interfaces
g
XGMII Extender
XGMII
XGXS
XGXS
8B/10B
8B/10B
XAUI
XGMII
Faster: 10X
Further: 40 km
n
Management: Consistent
Simple, Predictable, Elegant
Local
Clock
Line
Terminating
Equipment
(LTE)
Section
Regenerator
(STE)
Line
Terminating
Equipment
(LTE)
Stratum Clock
Note: A Line can be longer than two sections
http://grouper.ieee.org/groups/802/3/ae/public/terminology.pdf
Path
Terminating
Path
Equipment
Terminating
Path
(PTE)
Equipment
Terminating
Path
(PTE)
Equipment
Terminating
(PTE)
Equipment
(PTE)
Local
Clock
PCS Frame:
Viewed as 9 x 17280 Octets
PCS Frame = STS-192c Frame
576 octets
17280 octets
9 rows
Section
ort
p
s ad
n
a
Tr erhe
Ov
Line
9 rows
Section
rt
o
sp ad
n
Tra erhe
Ov
17280 octets
Line
Path Overhead
column
packet
IDLE
PCS data stream
Payload Capacity
IDLE
1
63
16640
16704 octets
9 rows
576 octets
Fixed Stuff
63 columns
J1
B3
C2
G1
Defined overhead
octets (F2, H4, Z3-5),
unused by
10GE WAN PHY
(set to zero)
...
9 rows
calculated
Does NOT
n
n
Test Patterns
g
Required Built in
n
n
n
n
n
n
Other
n
CRPAT
MAC
n
PHY
n
n
PMD
n
n
OAM Overview
g
OAM Layer
P802.3ae LAYERS
Higher Layers
OSI
Reference
Model
Layers
LLC
OAM
MAC Control
MAC
Application
Presentation
GMII
Session
Transport
Network
Data Link
Physical
MII
MII
PCS
PCS
PCS
PMA
PMA
PMA
PMD
PMD
PMD
MDI
MEDIUM
MDI
MEDIUM
MDI
MEDIUM
OAM Ping
Operation
g
Local
Remote
Client
Client
LLC
LLC
OAM
OAM
MAC CTRL
MAC CTRL
MAC
MAC
RS
RS
GMII
GMII
PCS
PCS
PMA
PMA
PMD
PMD
MDI
MEDIUM
Tx
Rx
MDI
MEDIUM
Tx
Rx
Local
Remote
Client
Client
LLC
LLC
OAM
OAM
MAC CTRL
MAC CTRL
MAC
MAC
RS
RS
GMII
GMII
PCS
PCS
PMA
PMA
PMD
PMD
MDI
MEDIUM
Tx
Rx
MDI
MEDIUM
Tx
Rx
Sample calculation:
n
n
n
30kb frame
acceptable probability 1%
BER 5 x 10 6
g
g
g
Status PDU
Event notification PDU
Variable request and response PDUs
n
Point-To-Point Overview
g
Based on FDDI
New
#
SMF MMF
Fibers (km) (m)
Tx
(nm)
Rx
(nm)
Rx
Sen
(dBm)
1000BASE-EX
>10
-20
1000BASE-BX-OLT
>10
1480-1500 1260-1360
-20
-BX-ONU
>10
1260-1360 1480-1500
-20
100BASE-LX
>10
1260-1360 1260-1360
-25
100BASE-BX-OLT
>10
1480-1580 1260-1360
-30
-ONU
>10
1260-1360 1480-1600
-30
1000BASE-PX-OLT-A
>10
1480-1500 1270-1360
-26
-ONU-A
>10
1270-1360 1480-1500
-25
-OLT-B
>20
1480-1500 1270-1360
-29
-ONU-B
>20
1270-1360 1480-1500
-25
To other
Subscribers
g
g
g
g
Bridge Tap
Transmission Characteristics
g
Attenuation
n
Crosstalk
Predominant impairment in loop plant
n Interference from same type of service on other
pairs in binder (self-crosstalk), or other types of
service (alien-crosstalk)
n
POTS/ISDN overlay
n
FEXT
Remote
PHYs at
subscriber
end
NEXT
Channel Capacity
g
N(f )
n
n
n
n
Regulatory Issues
g
Loop Unbundling
Loops in a binder may be operated by different
Telcos
n Crosstalk from pairs operated by one company will
affect performance on pairs operated by another
n
Spectral compatibility
Spectral limits and deployment guidelines to
ensure fair use of binder resources
n Mandated by national regulators (FCC, etc.)
n
ANSI T1.417
U.S. standard for spectral compatibility
n Requires demonstration of compatibility with
widely-deployed basis systems
n
Various DSLs
HDSL symmetric, T1 carriage, no POTS overlay
n ADSL asymmetric, POTS overlay, medium-long
loops
n VDSL symmetric & asymmetric, short loops,
high speed
n
EFM Copper:
Based on DSL Technologies
g
EFMC: An Evolutionary
Improvement over Existing DSL
g
Built to
accommodate
services that were
never deployed
Result is additional
costs for needless
provisioning,
configuration, and
maintenance
PPP
Unused
AAL5
Fast
ATM
path
Slow
ATM
path
Ethernet
Current typical
DSL protocol
stack a byzantine
collection
Mux
PMD
Typical DSL Modem
To PC
or
gateway
DSL
Modem
IP
Ethernet
DSLAM
PPP
AAL5
Fast
ATM
path
Ethernet
ATM
SDH
SDH
Fiber
xTU-C
Copper Loop
AAL5
Slow
ATM
path
Ethernet
Router
PC
IP
Ethernet
Mux
PMD
Ethernet
PHY
PHY
Cat5
omahony_copper_1_0702.pdf as the
basis for the line code evaluation criteria.
limit proposals for consideration regarding
the long reach objective to those based on
artman_copper_1_0702.pdf and
jackson_copper_1_0702.pdf
EPON Overview
Point-to-multipoint fiber network
High bandwidth: 1 Gbps shared
Low cost Ethernet + low cost fiber plant
Minimizes use of fiber, CO feeders, and
transceivers
Passive optical infrastructure
Fiber-to-the-home/building/business
applications
Suitable for voice, data, and video services
Point-to-Point Ethernet
!
!
N or 2N fibers
2N optical transceivers
32 or 64 fibers
64 transceivers
P2P
1 trunk fiber
Minimum fiber/space in CO
2N+2 optical transceivers
Electrical power in the field
1 or 2 fibers
66 transceivers
P2P
1 trunk fiber
Minimum fibers/space in CO
N+1 optical transceivers
No electrical power in field
Drop throughput up to trunk rate
Downstream broadcast (video)
curb switch
1 fiber
33 transceivers
P2MP
passive
optical splitter
EPON Configuration
1:N optical
splitter
single fiber
optics
VoIP
GW
Service
Providers
Video
Server
Model 2
L3
Service Multiplexing Switch
L3
L2
SP1
IP
Router
SP2
VoIP
GW
SP3
Video
Server
SP4
P2M P
OLT
Demarcation
Point
P2M P
ONT
P2P
OLT
P2P
OLT
P2P
OLT
Network
Operator
P2P
ONT
P2P
ONT
OLT
EPON Downstream
Physical broadcast of 802.3 Frames
Payload
ONU 1
FCS
1
USER 1
802.3 frame
3
1
1 3
1 3
OLT
2
ONU 2
USER 2
1
3
1
2
3
ONU 3
USER 3
OLT
MAC Control Client
Start
Stop
Clock register
Timestamp
GATE message
MAC
PHY
TS
Start
Stop
I N
MAC Control
MA_CONTROL.indication(GATE)
S C O P E
MA_CONTROL.request(GATE)
MA_DATA.request( )
MAC Control
TS
Write registers
Start
Stop
Clock register
Slot Start register
Slot Stop register
Laser ON/OFF
MAC
PHY
Upstream
Data Path
Generate GATE
message
EPON Upstream
ONU 1
USER 1
No packet fragmentation
1
1
OLT
ONU 2
1 1
3 3 3
USER 2
time slot
802.3 frame
ONU 3
USER 3
3
header
Payload
FCS
OLT
MAC Control Client
Clock register
TS
TBD
RTT register
Measure Round-Trip Time
MAC
PHY
TBD
I N
MAC Control
MA_CONTROL.request(REPORT)
S C O P E
MA_CONTROL.indication(REPORT)
Generate REPORT
message
MAC Control
TS
TBD
TBD
MAC
PHY
Clock register
Timestamp
REPORT message
2.
OLT sends
GATE at T1
ONU receives
GATE and sets
its clock to T1
T1
T3
(T3-T1)
GATE
OLT
Tx
T1
...
Rx
T2
4.
ONU sends
REPORT at T2
OLT receives
REPORT at T3
OLT calculates
RTT = T3 T2
***
REPORT
ONU
Tx
T2
Rx
T1
...
...
GATE
T1
5.
...
REPORT
(T2-T1)
3.
**
T2
Hybrid Fiber/Copper
Next-generation, high-speed architectures
EFM copper for the last 700 to 800 meters
Minimum 10 Mbps higher if possible
High bandwidth for entertainment client/server
For stepwise buildout to work, EFMCu must support next-gen
applications
P2P EFMF 1000 or 100 Mbps
P2MP EPON 32 Mbps per ONU
06/03/2002
Source: EFMA 2002
1.E+10
1.E+09
1.E+08
1.E+07
1.E+06
1.E+05
1.E+04
1.E+03
1980
1985
1990
1995
2000
2005
Business Models
Customer Usage
Wireless
Regulation
Competition
Free Space Optics
Economics
Technology
Applications
FSAN
Ethernet
Culture
CHAOS
xDSL
APON
Fiber
This chaos cannot be resolved by some central authority
IEEE 802.17
aka Resilient Packet Ring
aka RPR
aka ?Ethernet Loop?
RPR Overview
g
g
g
g
g
g
Inner Ringlet
Station
Station
Span
Station
Station
Station
TTL
2 Octets
RPR HEADER
6 Octets
6 Octets
2 Octets
m Octets
PAYLOAD
4 Octets
FCS
MODE
RI
PRI
IOP
TTL(8 bits)
n
n
Time To Live
Set to number of hops
to destination
Decremented when
forwarded by node
Allows for 255 nodes
on ring
MODE(3 bits)
n
Frame type
Mode
Value
Description
Reserved
Reserved
Reserved
Protection Control
Control
Fairness
Data
Origination ringlet
Value Description
Clockwise ringlet
Counterclockwise ringlet
Out of profile
In profile
High priority
0-6
Low priority
Cntl
Ringlet Input
MFU of
mate
MAC Fairness
Control Unit
HP
Drop
Logic
LP/MP
add
STOP_LP/MP/HP
MAC Client
Ringlet Output
Policer/
Shaper
M : rate monitor
Reserved (A0)
n
High (A1)
n
n
Medium (B)
n
n
n
Low (C)
n
n
Best effort
Subject to RPR-FA
Forwarding
n
Policing
n
n
n
n
Dynamic shaping
n
Access Rules
HP Transit
Has Frame?
HP Transit
LP Transit
Nearly Full?
LP Transit
Add HP/MP?
P
HP/MP Add
Trigger Flow
Control
Add MP/LP?
FA P
LP Transit
Has Frame?
Update FA
eMP/LP Add
LP Transit
: Policing Engine
FA : Fairness Algorithm
g
g
Determine congestion
n
n
n
Ring Protection
g
Steering
n
n
n
n
Wrapping
n
n
Optional in standard
All traffic is wrapped around when a station detects a
failure in its neighbor
Fewer dropped frames
Steering Example A to C
Station
C
Station
D
Station
B
Station
A
Station
F
Station
E
Wrapping Example A to C
Station
C
Station
D
Station
B
Station
A
Station
F
Station
E
Physical Layer
g
g
Campus Ring
MAN
Access Ring
Access
Ringlet
Backbone
Backbone
Fair Comparison?
RPR SONET Ethernet FCAL**
Fair access to ring bandwidth
!?
!?
!?
!?
But, are rings the way of the future, or a simply a means to replace SONET in the metro?
RPR Conclusion
like Ethernet
Frame based
Transceivers, Fibers,
and Issues with Optics
Typical
1 Gig Link
Protocol
Proposed
10 Gig Link
Protocol
Presentation
Coding
Session
SERDES
Coding
Transport
SERDES
Network
SERDES
XCVR
Data Link
XCVR
Media
Physical
Media
XGMII
4 x 8 bit
4 x 10 bit
XAUI
4 x 1 bit
MDI
1 bit or
4 x 1 bit or
PCS
PMA
PMD
MAC
PCS
PMA
PMD
20
10
Protocol with
Integrated SERDES
GLM
10B
1d
1x9
GBIC
1d
1x9
GBIC
SFF
SFP
circa 01
Pluggable
1997
1996
1x9
GBIC
SFP
SFF
2001
2002
XENPAK
FTRX
Seen at Optical Fiber Conference:
XENPACK; FTRX (300 pin MSA)
XXP; XPAK; XFP; SFP (@10 Gig!)
XGXS
Single mode
Fiber
62.5 m
125 m
Challenge:
Control
mechanical
tolerances
over
temperature
9 m
125 m
The vast majority of the cost difference is in the size of the target!
g Bandwidth
n
* Distance Product
Modal Dispersion
n
g Chromatic
Dispersion
Fiber Attenuation
More Loss
Attenuation (loss per meter)
Early 1970s
Mid 1970s
Early 1980s
Less Loss
600
800
1000
1200
1400
Wavelength in nanometers
1600
1800
Effects of Dispersion
Optical power at fiber input
850 nm Oxide VCSEL @ 1.25 GBd
Modal Dispersion
g
The smaller the core of the fiber, the fewer the number
of modes that will propagate
Single mode fiber (SMF) has only one mode and
therefore no modal dispersion (e.g., railroad)
Multi-mode fiber (MMF) profiles are doped so that all
paths take about the same time. Index at center of fiber
slows down low order modes
Chromatic Dispersion
Spectral width ()
) is measure
of the sources color range
Chromatic dispersion is
reduced by controlling the
sources
n
1310 nm
Velocity
1550 nm
Limit 2: Receiver
Dynamic Range =
Maximum - Minimum
optical power into
receiver
Rx min ( d B m)
Tx Min ( d B m)
Budget (dB)
Fib e r A llo c .
A ttn ( d B /km)
Dis ta n c e ( km)
Rx /Tx ma x
Dy n a mic Rn g
g
8 0 2 .3 z
-19
-11
8
2 .5
0 .5
5
-3
16
Ne w FC
-20
- 9 .5
1 0 .5
5
0 .5
10
-3
17
How Is 20 to 50 km
Achieved with 1300 nm LW?
g
Increase the
sensitivity of the
receiver (APD)
Increase the
dynamic range of
the receiver
Rx min ( d B m)
Tx Min ( d B m)
Budget (dB)
Fib e r A llo c .
A ttn ( d B /km)
Dis ta n c e ( km)
Rx /Tx ma x
Dy n a mic Rn g
8 0 2 .3 z
-19
-11
8
2 .5
0 .5
5
-3
16
O th e r
-22
0
22
20?
0 .5 ?
40
>2
>2 4
More Rx sensitivity
More Rx dynamic
range or engineer
link to bound
attenuation
Control the :
8 0 2 .3 z
Rx min ( d B m)
Tx Min ( d B m)
Budget (dB)
Fib e r A llo c .
A ttn ( d B /km)
Dis ta n c e ( km)
Rx /Tx ma x
Dy n a mic Rn g
O th e r
-32
0
32
25?
0 .2 5 ?
100
>1
>3 3
g
g
n
n
Robust, difficult to
create and validate
850 nm OMA
1550 nm Tx speced
at TP3 (chirp)
This scheme is
still used for
10GBASE-LX4
10GBASE-LX4 Tx Specifications
10GBASE-LX4 Rx Specifications
10GBASE-L Tx Specifications
10GBASE-L Rx Specifications
The Challenge:
Putting Down the Fiber
Fiber Recommendations
Outside the building? Install SMF
n
Longer distances
Potential for upgrade to DWDM
Inside building
n
n
Infrastructure Issues
g
n
n
Micro Trenching
Blown Fiber
Sin to not be filling open ditches with conduit (if not fiber)
Distribution Costs
fiber to hub materials
8%
8%
2%
1%
50%
6/18/02
Distribution Costs
Distribution of Home Capital
3% 5%
26%
7%
0%
1%
21%
0%
0%
26%
Budgetary Pricing
Labor
Connectivity
Corning, June 02
Cable Assemblies
Budgetary Pricing
Labor
$Feeder
Fiber Cable
FOHW
Corning, June 02
Distribution
Splice Closures
Connectivity
Drop
Cable Assemblies
Street Cutting
Excavation
TFS
Confidential & Proprietary.
Temporary Restoration
TFS
Confidential & Proprietary.
Micro-Trench
Up to 4 Cables per
Cut
Low Intensity
Construction
Non-Destructive
Installation
Rapid Deployment
Improved Agency
Acceptance
TFS
Confidential & Proprietary.
MTC Technique
Cutting the Micro-Trench
Shallow Depth Trench
Narrow Width Cut--10mm
Fully protected in
Hardened Space
TFS
Confidential & Proprietary.
MTC Technique
(cont)
Why MTC?
Traditional Carrier Class Depth Cost Prohibitive
to Address Last Mile Development
Other Alternatives (Sewer/Gas lines) Too Complex
for Wide Adoption
Match Solution to Application
TFS
Confidential & Proprietary.
TFS
Confidential & Proprietary.
MTC
Less Than
All Other
Options
MTC Advantages
Fastest Fiber Deployment/Delivery Method
Available Permitting Through Construction
Delivers Access and Point-to-Point Fiber Solutions
Delivers Fiber At Wire Line Prices
Minimal Disruption To Pedestrian and Traffic Flow
Survivable and Diverse Entry Topologies
Very Rapid Repair and/or Restoration
TFS
Confidential & Proprietary.
The Concept
Fibre unit bundles are then blown down the tubes on demand.
When your customers ask for a connection, small optical fibre units are
blown into the micro-tubes, without the need to splice.
Branching can be done anywhere along the route by cutting into the
protective jacket and connecting the existing micro-tube to a branch
micro-tube using a permanent or push/pull connector.
The Fibreflow solution can provide fibre optic links all along the network
on a Just in time basis
Sales Generation
Profit Generation
Savings
A
1,200,000
1,000,000
800,000
600,000
400,000
200,000
0
Fibre Splices
Fibre
Trench
Traditional
Fibreflow
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must Be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
Ptolemaic System
Ptolmys Epicycles
And then.
Portrait of Copernicus
Before 1584 AD - Tobias Stimmer
Network 3
Logical Link 2
Physical 1
Complexity Resolved
Layer 2
Logical Link
Ethernet
Token Bus
Token Ring
Modems
ISDN
SONET
FDDI
DQDB
ATM
PPP
SMDS
Frame Relay
VIP
IP
IPX
DDP
DLSw
IDP
COFP
CLNP
DRP
Convergence == Simplicity
Application 7
Presentation 6
Session 5
Transport 4
Network 3
Logical Link 2
Physical 1
In the future,
network
market
segments will
not be defined
strictly by
geography
Ethernet Trucks
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
SONET Ferry
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Packet
Business Models
Customer Usage
Wireless
Regulation
Competition
Free Space Optics
Economics
Technology
Applications
FSAN
Ethernet
Culture
CHAOS
xDSL
APON
Fiber
This chaos cannot be resolved by some central authority
Licensable
Open Standard
Other Proprietary
CSIX
SONET
InfiniBand
ATM
Ethernet
3GIO
Fibre Channel
SP
LDT
SP-2
RapidIO
PCI
FSB
HL-1&2
Processor Bus
Back plane
Storage/System
Area Network
Local Area
Network
MAN/WAN
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must Be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
10 to 1 Gig Price-Performance
Component
Ratio
System
Optics
SERDES
40 : 1
NIC
20 to 30 : 1 10GBASE-LR : 1000BASE-LX
40 : 1
10 : 1
10GBASE-LR : 1000BASE-LX
1000BASE-LX (seemed
unreasonably high)
10 GbE Price/Performance
n
$6,000
$6,000
$5,000
$5,000
$4,000
$4,000
10GigE actual
$3,000
$3,000
$2,000
$2,000
$1,000
$1,000
10 GbE
$0
2000
2000
2001
2001
2002
2002
2003
2003
2004
2004
10 GbE Price/Performance
n
$50,000
$50,000
10GigE actual
$0
2000
2000
2001
2001
2002
2002
2004
2004
Optical
Early: 850 nm CD Lasers; 1310 nm FP Lasers
n Mature: 850 nm VCSEL Lasers; 1310 nm FP Lasers
(1310 nm VCSELs soon?)
n
Packages
Early: OLM
n Mature: SFP; Integrated MAC/PHY/SERDES
n
Optical
Now: 850 nm VCSEL Lasers; 1310 & 1550 nm DFB
Lasers
n Future: 850 & 1310 nm VCSEL Lasers; 1550 ?
n
Packages
Now: XENPAK (XAUI); FTRX (300 pin)
n Future: {XXP; XPAK; XFP; SFP}?
n
HSSG
FORMED
STD!
WG
Ballot
LMSC
Ballot
TF
Review
PAR Approved
Preliminary
1 Gig Optics
Available
N
O
V
J
A
N
M
A
R
M
A
Y
J
U
L
Y
S
E
P
T
N
O
V
J
A
N
M
A
R
M
A
Y
Draft 4.2
Draft 4.3
Draft 5
S
E
P
T
Draft 4.1
J
U
L
Y
Draft 4
M
A
Y
2002
Draft 3.3
M
A
R
Draft 3.2
J
A
N
Draft 3.1
N
O
V
Draft 3
S
E
P
T
Draft 2.1
J
U
L
Y
Draft 2
M
A
Y
2001
Draft 1.1
M
A
R
2000
Draft 1
1999
J
U
L
Y
Preliminary
10 Gig Optics
Available
1 Gig
Interoperability
10 Gig
Interoperability
La
w
M
oo
re
40 Gig Next?
100000
2 0 06
2 0 02
10 Gigabit
GigabitEthernet
Ethernet
10000
1 9 98
Gigabit
Ethernet
Gigabit Ethernet
1000
1 9 95
Fast Ethernet
100
1 9 83
10
1980
1 9 90
1985
1990
10BASE-T
10BASE-T
1995
2000
2005
2010
La
w
M
oo
re
100000
2006
2002
10 Gigabit
GigabitEthernet
Ethernet
10000
1998
Gigabit
Gigabit Ethernet
Ethernet
1000
1995
100
Fast Ethernet
1983
10
1980
1990
1985
1990
10BASE-T
10BASE-T
1995
2000
2005
2010
Reasons for 40
g
Bits per
Hertz
Hz
1.00
8G
0.90
0.80
7G
100Mbit
10Mbit
0.70
6G
Gigabit
0.60
0.50
5G
0.40
3G
4G
0.30
2G
0.20
0.10
1G
0
0.00
1991
1993
1995
1997
1999
2001
2003
2005
2007
11
Inter-Board
Blade-to-Blade
Intra-Cabinet
Inter-Cabinet
Rack-to-Rack
Box-to-Box
Inter-Facility, Enterprise,
Site-to-Site, VSR,
Data Center-to-Data Center
20
15 m
30 m
300 m
Source: Intel
10
5
Third Generation
I/O Architecture
1
1GHz Parallel Bus Limit
UP TO 800 MHz
HL
PCIx
HT
AGPx
UP TO 66 MHz
VL
8.33 MHz
EISA
ISA
80s
PCI
Max Bandwidth/Pin
Flexibility
Multiple market segment
MCA
90s
Point-to-point
R I/O
VESA
Full Serial
00s
Source: Intel
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must Be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
Did I mention
Ethernet in the First Mile?
Backbone Glut
or Access Dearth?
Metro
Continent-to-Continent
Coast-to-Coast
all over Fiber at 10 Gbps & up
City-to-City Town-to-Town
all over Fiber at 1Gbps 10 Gbps
Access
LAN
Desktop-to-Desktop Floor-to-Floor
10 Mbps 1Gbps
10GbE
(802.3ae)
10GbE
(802.3ae)
1000BaseLX
1300nm
9 SMF
1000BaseLX
1300nm
1000BaseSX
850nm
1000BaseSX
850nm
10GbE
(802.3ae)
10GbE
(802.3ae)
LX4 1310nm WWDM; 300m on 62.5 MMF, 500 MHz*km; LAN PHY only
1000BaseSX
850nm
1000BaseSX
850nm
1000BaseT
(802.3ab)
4 Pr Cat5 UTP
1000BaseCX
Copper
Balanced
Copper
25m
100m
65m
220m 275m
300m
500m
550m
5km
10km
40km
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must Be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
Page - 9
WWW.TCPIPSERVICES.COM
2002 IP Services - All Rights Reserved
Risk Assessment
Rollback and Contingency Mgmt.
Plans for Actual Release Roll-outs
Incident Management
Escalation Management
Problem Management
SLA (Service Level Agreement)
Management
OLA (Operation Level Agreement)
Management
Reporting on Actual Performance
vs. Contract Terms
Asset and Configuration Mgmt.
Change Management
Monitor and Maintain Configuration
Baselines
What Is an OSS/BSS
0018-02
Arent ATM/SONET/SDH
Better Than Ethernet for QoS?
This is the Wrong Question
Get over it !
Right question
n
Optical Ethernet
Deficiencies and Mitigation
g
Providing QoS
Mitigation
n
Providing QoS
n
Mitigation
n
n
Mitigation
n
n
Single-ended maintenance
Loopback testing
Flow-through provisioning
Integrated operations support systems
Capacity planning and management
Service level agreements
Mitigation
n
n
n
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must Be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
Household Budget
Minutes
Hours
Modem 56 kbps
Days
13 days
Pony Express
11 days**
5 days
11 hrs 36 min
11 hrs 12 min
10 hrs
2 hrs 12 min
53.6 min
19.4 Mbps
30 min***
35 mph
1000 Mbps
10.4 min
1 min
* The Matrix DVD 7.18 GB from New York, NY 10005 delivered to Beverly Hills, CA 90210
** extrapolated from record: 7 days 17 hrs - approx 2,000 miles from St. Joseph, Missouri
to Sacramento, California Lincoln's Inaugural Address, March 4, 1861
*** if you live close no traffic its in stock & theres no line
$44
$110+ Monthly
Monthly
Revenue
Margin
Residential Subscriber
Cost of fiber plant $ 1,000 amortized* @ 20 years = $ 9 per month
Cost of electronics $ 1,500 amortized* @ 7 years = $ 25 per month
Cost of delivering content per subscriber = $ 32 per month
* Levelized cost at 8.5%
75 Voice lines
10,000 LD minutes
10 Mb Internet access
10 Mb LAN interconnection
Videoconferencing
Telemedicine
Voice
Data
$6,450
Monthly
Revenue
Payoff
~1
year
Business Video
Capital Costs
Cost of Lateral & Building Entrance Fiber = $ 50,000
Cost of Third-Party voice-switching equipment = $ 23,785
Cost of Ethernet Access 2 Gbps = $ x,xxx
Total = $ xx,xxx
6/18/02
Construction Summary
June 7, 2002
6/18/02
Economic Development
June 7, 2002
17 NOC Employees
28 other PUD Support
25 contract labor (3-5 yr)
2 NCESD, K20
>$16M Economic
Benefit
96 new jobs!
6/18/02
10
Lessons Learned
16
Summary
Grant County PUD FTTH Project
will influence community change by:
17
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must Be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
Distractions or Complements?
EtherEveryThings
n
n
n
n
Chip-to-Chip Communication?
60-90 GHz Pt-to-Pt Radio?
Ethernet Disk Drives?
Subspace?
EtherKin
n
n
Other
n
n
n
n
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
Regulatory Impact
The single, most profound
influence on the future of
networking will be the
acceptance and adoption of the
OPEN ACCESS MODEL
(or NOT)
Jonathan Thatcher; 2/2/2000 :-)
Reed E. Hundt
Customer as Hostage
Satellite
Wireless
ILEC
Cable 2
Cable 1
ISP
CLEC
Whos
Monopoly
Is It
Anyway?
g
g
g
Water
Roads
Sewers
T
E
L
C
O
s
Community
New Paradigm
Community Access Network
Users
Service Providers
Management
Concentration
Access Portal
Customer Choice
Satellite 3
Satellite 2
Satellite 1
Wireless 2
Wireless 1
Shared Infrastructure
Cable 4
Cable 3
Cable 2
Cable 1
ISP 1
ISP 2
ISP 3
ISP 4
CLEC 1
CLEC 2
CLEC 3
CLEC 4
ILEC
Community Access
Network
Operations Center
Clash of Paradigms
The Public Network at Bay
g
g
g
g
g
g
20th Century
Circuit switched
Centralized
Voice driven
Value in metering
use
Deterministic
Monopoly
g
g
g
g
g
g
21st Century
Packet switched
Decentralized
Data driven
Value apps and
services
Evolutionary
Competitive
Towards Simplification
Towards Higher Speed; Lower Cost vs.
Moores Law
Ethernet to the Rescue in the Access Space
QoS and OAM Can Be and Must Be Solved
Economic Models Can Support True
Broadband Services
Distractions or Complements
Federal Regulation and Policy Will Be the
Single Greatest Influence on Technology
Development
Investment as a Positive Feedback System
~
~
10 Mbps
SDTV (1)
5 Mbps
Gaming 2 Mbps
Web surfing 56 Kbps
Voice - 6.5 Kbps
Dial-Up
ISDN
56 Kbps 128 Kbps
Cable Modem
ADSL
1.5 Mbps
8.5 Mbps
not drawn to scale
PON
19.5 Mbps
Gigabit Ethernet
1,000 Mbps
Ethernet
# of Customers
100K
100K
Services
V-V-D
V-V-D
Install Cost
1.2x
Operational Costs
.5x
Equipment Life
2-3 years
Bandwidth
< 10 Mbps?
1.0 Gbps
Infrastructure Life
< 3 years?
>30 years
Company Value
400
300
110
200
10
100
30
0
2002
2003
2004
2005
2006
2003
2004
2005
2006
2002
2003
2004
2005
2006
Manufacturer
Revenue ($M)
Port
Shipments
(000s)
Manufacturer
ASP ($)
2002
2003
2004
2005
2006
250
525
1058.8
2290.8
3537.2
10
30
110
340
750
25,000
17,500
9,625
6,738
4,716
g
g
g
Related Organizations
& Technologies
OAM
Fiber Point-to-Point
Fiber PON
Copper
EFMA
g
g
g
g
An industry alliance
Support the standards
process with resources
Market the technology
Host interoperability
events
Proven concept
EFMA Goals
g
Marketing Goals
n
Technical Goals
n
Promotion Material
n
Technical Meetings
n
Delivering the
message
Participate in Events
n
Speakers Bureau
n
Technical
EFM Tutorials
n
Broaden
understanding
Inter-op Events
n
Prove products
interwork
10 GEA Mission
g
What Is OIF?
g
g
g
Request standardization
of specific OIF recommendations
Provide informational documents
to the target standards group
ANSI T1
IETF
ATM Forum
IEEE 802.3ae 10 Gbit Ethernet
NPF
Architecture
n
Network management
Interoperability
n
Signaling
n
Carrier
n
Interoperability testing
g
g
g
g
10G
10G
Framer
Framer
10G
10G
Optics
Optics
10G
10G
10G
10G
Optics
Optics Framer
Framer
16:12
16:12
16:12
16:12
C
C
O
O
N
N
VV
EE
R
R
TT
EE
R
R
C
C
O
O
N
N
VV
EE
R
R
TT
EE
R
R
1.24Gbps
Optics
VCSEL Array
Low speed
parallel links
Parallel Ribbon Cable
(<400m) 62.5um MM
10G
10G
Framer
Framer
1.24Gbps
Optics
Pin-PD Array
OIF Summary
g
MBytes/per sec.
25000
20000
15000
10GFC
10000
1GFC
5000
1.06
Gbaud
1x
2GFC
4GFC
2.12
Gbaud
2x
4.25
Gbaud
4x
12.75
Gbaud
12x
0
1996
2000
2002
2002
<2010
Adoption
Length
<300 meters
mostly <100 meters
<$500/GBIC (max!)
mostly <$100 (10G
will bear some
premium, but not
much)
P e r P or t C ost ($)
2001
2002
2003
2004
2005
Integrated 4-lane CMOS 10G Copper (FC and IB Only), 2002 <$20/port
4G FC Disks 1-lane CMOS Copper, 2003 <$10/port No plans for 4G Optical xvr!
Resistance Is Futile
Ethernet
Sh
ed
ar
M
ed
10
10 Gigabi
Gigabitt
Ethernet
Ethernet
ia
L
S
AD les
Ru
Limi
Broa t
Preserve our Copper
dban
d Save DH
N
o
ms
S
e
/
d
o
T
M
Vo
NE
er
O
IP
S
Forev