Download Design of an AAPN

Design of an
Agile All-Photonic Network (AAPN)
Gregor v. Bochmann
School of Information Technology and Engineering (SITE)
University of Ottawa
Canada
http://www.site.uottawa.ca/~bochmann/talks/AAPN-results
Presentation at the APOC
conference
Wuhan, November 2007
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
1
Abstract

Agile All-Photonic Networks (AAPN) is a Canadian research
network (funded by NSERC and 6 industrial partners)
exploring the use of very fast photonic switching for
building optical networks that allow the sharing
(multiplexing) of a wavelength between different
information flows. The aim is to bring photonic technology
close to the end-user in the residential or office
environment. The talk gives an overview of the proposed
overlaid star network architecture and describes new
results on (a) bandwidth allocation algorithms, (b) the
routing and protection of MPLS flows over an AAPN using
the concept of OSPF areas, and (c) our evolving plans for
building demonstration prototypes.
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
2
Overview





Overview of the AAPN project
Frame-by-frame bandwidth allocation
MPLS over AAPN
A demonstration prototype
Conclusions
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
3
Different forms of “burst switching“





Question: Can one do packet switching in the optical
domain (without oeo conversion)?
At a switching speed of 1 μs, one could switch bursts of
10 μs length (typically containing many packets)
Traditional packet switching involves packet buffering in
the switching nodes. Should one introduce optical buffers
in the form of delay lines?
The term “burst switching“ originally meant “no
buffering”: in case of conflict for an output port, one of
the incoming bursts would be dropped.
Note: Burst switching allows to share the large optical
bandwidth among several virtual connections.
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
4
AAPN
An NSERC
Research Network
The Agile
All-Photonic Network
Project leader: David Plant, McGill University
Theme 1: Network architectures
Gregor v. Bochmann, University of Ottawa
Theme 2: Device technologies for transmission and
switching
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
5
AAPN Professors






(Theme 1 in red)
McGill: Lawrence Chen, Mark Coats, Andrew Kirk, Lorne
Mason, David Plant (Theme #2 Lead), and Richard Vickers
U. of Ottawa: Xiaoyi Bao, Gregor Bochmann (Theme #1
Lead), Trevor Hall, and Oliver Yang
U. of Toronto: Stewart Aitchison and Ted Sargent
McMaster: Wei-Ping Huang
Queens: John Cartledge (Theme #3 Lead)
Note: Theme 2 deals with device technologies for
transmission and switching
For further information see: http://www.aapn.mcgill.ca/
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
6
The AAPN research network


Our vision: Connectivity “at the end of the
street” to a dynamically reconfigurable
photonic network that supports high
bandwidth telecommunication services.
Technical approach:


Simplified network architecture (overlaid stars)
Specific version of burst switching

Fixed burst size, coordinated switching at core node for all input
ports (this requires precise synchronization between edge nodes
and the core)


See for instance
http://beethoven.site.uottawa.ca/dsrg/PublicDocuments/Publications/Hall05a.pdf
Burst switching with reservation per flow (virtual connection),
either fixed or dynamically varying

See for instance
http://beethoven.site.uottawa.ca/dsrg/PublicDocuments/Publications/Agus05a.pdf
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
7
Agile All-Photonic Network
Edge node with slotted transmission
(e.g. 10 Gb/s capacity per wavelength)
Fast photonic core switch
(one space switch per wavelength)
Opto-electronic interface
- Provisions submultiples of a
wavelength
- Large number of
edge nodes
Overlaid stars architecture
Future of Networking, Lausanne, 2005
8
Starting Assumptions

Avoid difficult technologies such as









Wavelength conversion
Optical memory
Optical packet header recognition and replacement
Current state of the art for data rates, channel spacing,
and optical bandwidth (e.g. 10 Gbps)
Simplified topology based on overlaid stars
Large number of simple edge nodes (e.g. 1000)
Fixed transmission slot length (e.g. 10 msec)
No distinction between long-haul and metro networks
This requires


Fast optical space switching (<1 msec)
Fast compensation of transmission impairments (<1 msec)
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
9
AAPN Architecture

Overlaid stars
2
Core Node
1
Edge Node
A
3
6
B
4
8

Port sharing is required to allow
a core node to support large
numbers of edge nodes


7
5
A selector may therefore be
used between edge and core
nodes
A wavelength stack of bufferless
transparent photonic switches is
placed at the core nodes

a set of space switches, one
switch for each wavelength
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
10
Deployment aspects - Questions

Long-haul or Metro ?



connectivity “at the end of the street”; to a server farm
AANP as a backbone network ?
High capacity
(many wavelengths)
?
Multiple core nodes ?
or low capacity
(single or few wavelengths)




For reliability
For load sharing
Transmission infrastructure ?


Using dedicated fibers
Using wavelength channels provided by ROADM network
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
11
Overview


Overview of the AAPN project
Frame-by-frame bandwidth allocation
(work by my PhD student Cheng Peng)



MPLS over AAPN
A demonstration prototype
Conclusions
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
12
Comparing Burst-Mode Schemes


Long-haul AAPNs: long propagation delays
for signalling
Two modes of slot transmission:



Collaboration with Anna Agusti-Torra
(Barcelona)



With reservation (long signalling delay)
Without reservation, as proposed for “Burst Switching” (loss probability
due to collisions)
New method: Burst switching with retransmission (to avoid losses)
Comparison with TDM (see next slide)
Method to avoid long signaling delays: TDM

Allocate unused time slots; these free slots can be used without
signaling delays (they were allocated in advance)
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
13
TDM vs. OBS

OBS-R
What kind of
technologies should be
employed in the AAPN,
TDM or OBS?


TDM

Gregor v. Bochmann, University of Ottawa
The delay of OBS w/
retransmission (OBS-R)
degrades sharply when
the load is beyond 0.6
but is negligible at
lower load.
The delay of TDM
maintains better delay
performance at the
high load compared
with OBS-R.
TDM shows a better
performance than
OBS-R especially at
the high load.
AAPN, 2007
14
Birkhoff - von Neumann Approach


The BvN decomposition approach
calculates the timeslot schedules for a
frame from the traffic demands between
all node pairs.
Two steps:



Constructing a service matrix from a traffic matrix
Decomposing the service matrix into switch permutations.
(problem has O(N4.5) complexity)
The main challenges of BvN
Decomposition are:
How to construct a service matrix that closely reflects
the traffic demand for all source-destination pairs?
 How to find a heuristic decomposition algorithm with low
complexity that allows a practical implementation?
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
15

Service matrix construction

similarity of final service matrix
b)
0.95

0.9
projection method,  = 0.5
projection method,  = 0.25
max-min fairness
0.85
20
60
40
N
Gregor v. Bochmann, University of Ottawa
80
New algorithm:
Alternating
Projections Method
Similarity Comparison
 Compared with Max-min
fairness method [7]
 The service matrices
obtained with this
Alternating Projection
method have very high
measures of similarity to
the original traffic matrix,
with an average similarity
greater than 95% for
N>=32.
AAPN, 2007
16
Mean Queueing Delay (timeslot)
Service matrix construction: queuing delay

b)
8000
7000
Delay performance

Alternating Projections Method
Simple Rescaling Method
Max-Min Fairness Method

6000

5000
Long-haul scenario,
N=16, 1000km
Tested under selfsimilar traffic
Compared with

4000

3000
2000

1000
0
0.1
0.2
0.3
0.4
0.5
0.6
Offered Load
Gregor v. Bochmann, University of Ottawa
0.7
0.8
0.9
Max-min fairness
method [7]
Simple rescaling
method [8]
Conclusion:
performs better than
the max-min fairness
method or the simple
rescaling method.
AAPN, 2007
17
Mean Queueing Delay (timeslot)
Heuristic decomposition algorithm
b)
4

10
QBvN
EXACT
GLJD
New algorithm:
Quick BvN
(QBvN)


3
10

Long-haul scenario,
N=16, 1000km
Tested under self-similar
traffic
Compared with


2
10
0.1
0.2
0.3
0.4
0.5
0.6
Offered Load
Gregor v. Bochmann, University of Ottawa
0.7
0.8
0.9 
Benchmark: Exact
BvN
Greedy Low Jitter
Decomposition (GLJD)
[11]
Conclusions:
 excellent performance
(close to optimum)
 Low complexity O(NF)
AAPN, 2007
18
Overview



Overview of the AAPN project
Frame-by-frame bandwidth allocation
MPLS over AAPN
(work by my PhD student Peng He)


A demonstration prototype
Conclusions
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
19
IP/MPLS routing over an AAPN
...
Edge node
Edge
node
router
Edge node
router
Edge
node
Core node
er
rout
Core node
router
Edge node
Edge node
...
Edge node
AAPN
Edge node
router
router
One OSPF Area
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
20
Solving the scalability problem
Applying OSPF in a straightforward manner over
an AAPN:


AAPN with N edge nodes corresponds to N x N links in
the routing table (1 000 000 links in case of 1000 edge nodes)
Different approaches to introduce abstraction (we
speak about “virtual routers” )
Vi
No rtua
de l
#2
...
...
Edge
node
Edge node
Edge node
Edge node
Edge node
Edge
node
Edge
node
the core
Edge
node
Edge
node
Edge node
Edge
node
Edge node
...
Edge
node
N
Edge node
Vi
od rtu
e al
#1
...
the core
Edge node
Edge
node
Edge node
Edge node
Edge node
the core
Edge
node
...
...
Edge node
Edge node
Edge node
Edge node
...
...
Edgenode
Edge node
(a) AAPN exported as a full-meshed topology
(c) AAPN exported as a edge-star topology
l
ua 4
rt #
Vi de
o
N
Edge node
Edge node
Edge node
Virtu
Nod al
e #3

Edge
node
(d) Virtual node organization
Edgenode
Edge node
(b) AAPN exported as a core-star topology
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
21
Result: Optimal OSPF inter-area routes

Area #2
OSPF can support multiple routing areas with only
local routing
Area #1
information
v-ABR
..
.
Finding optimal inter-area
research problem
v-ABR routes is an open
..
.
This can be solved when
the OSPF areas are interconnected by an
AAPN: MPLS over AAPN Traffic Engineering Framework
E4
R6
R2

E1
R8
E5

R3
R1
E2
E6
R7
R4
E3
R5
Area #1
Area #2
Area #0
R2
E1
E4
..
.
R3
R1
R6
..
.
E2
R8
E5
R4
E3
AAPN
Architecture
E6
R7
R5
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
22
Multi-layer resilience of MPLS flows over AAPN

Using our traffic engineering framework as a
starting point, we are working on multi-layer
resilience of MPLS flows over AAPN, especially
for inter-area (OSPF), intra-provider inter-AS and
inter-domain network environments.
End-to-end inter-area Resilience
IP/MPLS Resilience
IP/MPLS Resilience
AAPN
Optical
Resilience
Edge
Resilience
Edge
Resilience
Edge-to-edge Resilience
Area #1
Area #2
Area #0
Vertical: Multi-layer
R2
E1
..
.
R3
..
.
R8
E5
E2
R1
R6
E4
R4
R4
E3
AAPN
Architecture
E6
R7
R5
Horizontal: Multi-area
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
23
Internet Exchange architecture with TE

The principle of Virtual
Routers can be applied
to star-like networks in
other situations:


…
Ethernet
Switch
(edge)
Aggregation
Ethernet Switch
(core)
IST
SMLT
BGP instead of OSPF
Star-like Ethernet instead of
AAPN
…
ASBR

Examples:

Edge node
AAPN-based
IX (AIX)
Edge
node
ISP B
(AS y)
...
...
Core node
Edge node
Edge
node
ASBR
Edge node
Core node
…

existing Internet Exchange
based on Ethernet
Internet Exchange using an
AAPN
ASBR
ISP A
(AS x)
Core node
ASBR
Edge node
(E1)
Optical
Fiber
...
Edge node
ASBR
ISP C
(AS z)
...
Route
Server
Content
(e.g., IPTV)
Server
Edge node
ASBR
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
24
Overview




Overview of the AAPN project
Frame-by-frame bandwidth allocation
MPLS over AAPN
A demonstration prototype
(in collaboration with the whole Theme-1 team)

Conclusions
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
25
Building an AAPN Control Platform

Realizes algorithms and protocols for controlling an
AAPN




Easily adapting to the control interfaces of different core
switches
Easily integrating various control algorithms and protocols
developed by different AAPN Theme-1 researchers
Initial version may be running at slower speed (using standard
PCs with optical Ethernet cards)
Control information is exchanged through normal data blocks
(no separate control channel)
E1
E2
E3
E4
Co-located
M1
E5
M2
Core
Control interfaces
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
26
Node architecture
IP traffic layer
Global
SECM
(edge node
functions
specific to
bandwidth
allocation
algorithm
in core node)
GECM
(generic
signaling
protocol)
Switch control
interface
Gregor v. Bochmann, University of Ottawa
AAPN layer
Packet
aggregation
Burst buffering
and transmission
Burst reception
and packetization
Transmission & synchr. layer
AAPN, 2007
27
Example timing diagram
a frame period
signalling
Gregor v. Bochmann, University of Ottawa
corresponding frame period
at core node
corresponding data transfer
AAPN, 2007
28
- using essentially the same
control software

Slow prototype

Control Platform runs on PCs, optical Ethernet
transmission, MEM switch (slot time = 1 second)


Completed in January (used to test control platform software)
Intermediate prototype

Most of Control Platform runs on PCs, optical
transmission at 1 Gbps controlled by FPGAs, nanoseconds optical switch, slot time = 0.2 milliseconds


Expected to be completed in July
Improvements and extensions




Add basic MPLS transmission in the IP Traffic Layer to allow
experimentation with real applications
MPLS routing and traffic engineering (TE)
Experiment with bandwidth allocation algorithms proposed by
AAPN researchers
Other AAPN architectures (e.g. concentrators)
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
29
Component-based
edge node architecture
PC with
apps
LAN
IP / MPLS / Ethernet
switching / routing
+
AAPN routing
“basic” AAPN:
Slot transmission
Bandwidth alloc.
PC with
apps
e.g. el. Giga-Ethernet
PC with
apps
LAN
“basic” AAPN:
IP / MPLS / Ethernet
switching / routing
+
AAPN routing
Slot transmission
Core node 1
Bandwidth alloc.
PC with
apps
“basic” AAPN:
PC with
apps
LAN
PC with
apps
Slot transmission
IP / MPLS / Ethernet
switching / routing
+
AAPN routing
Core node 2
Bandwidth alloc.
AAPN edge node
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
30
Conclusions



The AAPN is a fascinating research project
Premise: very fast switching and simple
network architecture (overlaid stars)
Theme-1 Outcomes:



Theme-2 Outcomes:


Proposal for slotted burst switching with bandwidth reservation
(TDM, various allocation schemes)
Internet-integration : AAPN as an OSPF area
Various technologies for very fast switching, amplification, RRR,
etc.
Demonstration prototypes
Gregor v. Bochmann, University of Ottawa
AAPN, 2007
31