Download Document

Optical Packet Switching
the technology and its potential role in
future communication networks
Results from IST project
.
Lars Dittmann
COM – Technical University of Denmark
[email protected]
[email protected]
Zagreb 210503
1
What is the
next generation photonic network?
(different targets different timescales)
• Extension to current SDH/SONET network with
LCAS, ASON, GMPLS, GFP, etc. ?
• Bitrate and protocol transparent optical datapath
with electrical control and management ?
• All-optical network with optical control,
information processing and routing ?
[email protected]
Zagreb 210503
2
Why do we need the
next generation optical network ?
• for cost reduction reasons (cost reduction potential seems
larger for optics than for electronics)
• to increase network efficiency and utilisation
• for resource savings preserving network reliability and
availability
• for better network control for fast and efficient configuration
of connections (reduction of manual interventions)
• to increase network flexibility and responsiveness to dynamic
traffic demands/changes
• because an optical network is in line with a simplified core
structure with more complex and intelligent flow handling at
the edges (which was the original idea of the MPLS concept)
[email protected]
Zagreb 210503
3
Technological challenges
possible optical connection types
Network Resources
Circuit Switching
Packet and Burst Switching
Connection
Oriented
Virtual Connection
Oriented
Leased Lines
Virtual Leased Lines
& VPNs
ConnectionLess
Burst
Switching
[email protected]
Zagreb 210503
Packet
Switching
4
Technological challenges
packets vs. circuits
10s
Circuit
1s
Flow
Transfer Time
100ms
10ms
1ms
100ms
10ms
B
1ms
e
at
itr
Burst
s
ps ps
bp
G Gb Gb
5 10 40
2.
Packet
100ns
100
1k
10k
0.1M
1M
10M
0.1G
1G
Transaction Size (Bytes)
[email protected]
Zagreb 210503
5
Why packet switching
• Primarily a traffic engineering tool!
• Seen as the final goal for network
flexibility, however must be justified
• Packet based operation at application level
and transport level should not be mixed up!
• Potential new methods for network
resilience in packet based networks (path
set-up without resource reservation)
[email protected]
Zagreb 210503
6
Optical Burst and label Switching
potential step towards optical transport plane
Control plane interface
E
Labels in
Switch Control
(Label Swap)
In Fibre
l0, l1,..., ln
WDM
Dem ux 1
Add/Drop control
lc
l1
ln
E
Labels out
Switch control
Optical
Space
Add/Drop Switch
Drop Ports
O
lc
l1
ln
WDM
Mux 1
O
Out Fibre
l0, l1,..., ln
Add Ports
Electrical Domain
Burst Assembly/Disassembly
Buffering
Optical Domain
O/E Conversion
Data interfaces
IP Router
[email protected]
Zagreb 210503
7
• DAVID = Data and Voice Integration over DWDM
• A European research project
– Financially supported by the EU commission
– IST program
• Goals
– Develop concepts and technologies for future, optical networks
– Traffic engineering in packet-over-WDM based networks
– Control systems for optical networks
• Timeline
– Start July 2000, end October 2003
[email protected]
Zagreb 210503
8
DAVID Project partners
• Companies
• Universities
– Alcatel, CIT (F)
– Alcatel, SEL (D)
–
–
–
–
–
–
–
• Network operators
– BT (UK)
– TELENOR (N)
– TELEFONICA (E)
• Research centers
NTUA (G)
University of Bologna (I)
Politechnica de Torino (I)
LRI (F)
INT (F)
University of Essex (UK)
UPC (E)
– IMEC (B)
– COM (DK)
[email protected]
Zagreb 210503
9
Overall architecture
• Key components
–
–
–
–
OPADM – optical packet add / drop multiplexer MAN
Hub
WAN
Gateway
OPR – optical packet router
• Coverage
– MAN and WAN
• Control
– MPLS-based
[email protected]
Zagreb 210503
10
Hierarchical MPLS concept
• An MPLS based architecture for mixed-technology
networks
• Traffic optimized/conditioned between levels
• Levels of various granularity
Level
Electrical MPLS
Optical MPLS
Wavelength
routed
Bandwidth
granularity
Packets
Larger packets
Wavelengths
+ wavelength bands, fibers
[email protected]
Zagreb 210503
11
The
optical packet MAN
Hub
• Topology - interconnected physical DWDM rings
• Each physical ring -> several logical rings
• Ring nodes – OPADMs - provide
OPADM
OPADM
– Ring connectivity
– Legacy network interfaces
• Inter ring traffic controlled by a hub
λ0
λ1
[email protected]
Zagreb 210503
...
λν
12
The metro part: MAN
• Hub functionality:
– each timeunit Hub switches traffic between rings
– permutations to switch multislots between (logical) rings
– permutations to use are based on measurements of
"demand"
ring 1
ring 1
?
ring 2
ring 1
ring 2
ring 2
ring 3
ring 3
ring 4
ring 4
ring 3
ring 4
slot
1
2
3
1
3
3
...
2
3
4
3
2
1
...
3
4
1
2
1
2
...
4
1
2
4
4
4
...
ring permutation
[email protected]
Zagreb 210503
ring 3
ring 4
input ring
ring 1
ring 2
13
Packet formats
For synchronisation reasons fixed size packets
(at transport level) is preferable for small units (nano-micro sec).
Variable service units handled by sequence of fixed size packets.
[email protected]
Zagreb 210503
14
Ring node evolutionary steps
(cost vs. flexibility)
DAVID MAN with passive
OPADMs
“Now”
[email protected]
DAVID MAN with active
OPADMs
“Future”
Zagreb 210503
15
Administrative challanges
ASON vs. GMPLS/MPlS
• Apply the dynamic configuration of service
layers to the transport layer(s)
• Integrated control of layers in the network
• Optimised use of the individual layers
• Standard proposals from: IETF, ITU-T, OIF
• Protocol centric solutions (IETF) vs.
architectural centric solutions (ITU-T)
• Multi-layer resilience concepts
[email protected]
Zagreb 210503
16
Administrative challanges
level 2
(OMPLS)
level 3
(MPλS)
IETF : MPLS/MPlS
level 0
(MAN)
level 1
(EMPLS)
WAN
edge
router
MAN
Technology hierarchy
[email protected]
Zagreb 210503
17
When will the next generation of
photonic networks become a reality
• Significant effort needed to lower the cost and enable
OAM functions of optical components (higher integration
and automatic packaging)
• Better understanding of traffic and performance issues in
core and metro networks needed to evaluate cost and
reliability issues in current proposals.
• Gain consensus on administrative concepts and standard.
• Optical networks must become digital – 3R in all elements
as first process
[email protected]
Zagreb 210503
18
When will the next generation of
photonic networks become a reality
• Dynamic administration of pseudo optical
networks (SONET/SDH) in 2-4 years.
• All-optical networks functions in the data plane
obtainable in 5-10 years
• All-optical operation in all layers is not realistic
with current know technology (and might never
be)
[email protected]
Zagreb 210503
19
IST DAVID info @ david.com.dtu.dk
Public demo in October in relation to
PS´2003 (photonics in switching) in Paris
and
(NGPN deliverables @ www.ngni-core.net)
[email protected]
Zagreb 210503
20