TCOM 513 Optical Communications Networks

1
TCOM 513Optical Communications Networks

• Spring, 2007
• Thomas B. Fowler, Sc.D.
• Senior Principal Engineer
• Mitretek Systems

2
Topics for TCOM 513

• Week 1 Wave Division Multiplexing
• Week 2 Opto-electronic networks
• Week 3 Fiber optic system design
• Week 4 MPLS and Quality of Service
• Week 5 Optical control planes
• Week 6 The business of optical networking
  economics and finance
• Week 7 Future directions in optical networking

3
Resources

• www.sorrentonetworks.com/whitepapers.asp
• Get their IP over Optical presentation
• www.tellium.com/optical/presentations.html
• Get Convergence of IP and Optics
• Other presentations useful as well
• www.nanog.org/mtg-9905/mpls.html
• Right click and you can get the slides (Nortel)
• www.cellstream.com/prod08.htm
• Multiprotocol Label Switching
• Youll have to pay for this one 27.95
• www.itprc.com
• Info about various routing protocols

4
Resources (continued)

• www.cis.ohio-state.edu/jain/
• Tutorials and papers on various networking
  subjects from Raj Jain
• www.cisco.com/warp/public/503/2.html
• Cisco networking icons in various formats
• www.iec.org
• Download MPLS tutorial from Trillium

5
Topics

• Switching problem and label switching
• MPLS
• MPlS
• Current Network Problems
• Enhancing Internet Protocol (IP) Networks To
  Support A Variety of Applications
• Quality of Service (QoS) As A Solution
• Real-time Application Protocols
• Two Locations for QoS Access And Backbone
• Diffserv and QoS
• Cyber Security and QoS

6
Economic reality Carriers dilemma
7
How can carriers find new high-margin service
offerings?
8
Network realitySONET infrastructure
9
Network reality DWDM

• Most packet data networks are meshed

10
How to best marry these three
11
Fundamental conflicts

• Topology and technology
• Data networks on SONET and DWDM
• Some services still require SONET 50 msec
  restoration
• Economics
• Packet data networks are naturally resilient
• May not justify cost for SONET redundancy in
  order to collect lower revenue for best effort
  service
• Providers are looking for network to support
  voice, private line, data with same
  infrastructure

12
How to deal with problem and retain (or improve)
profitability

• Migrate to intelligent optical networking
• Offer new services
• Higher bandwidth services
• Optical VPNs Public services that act like
  private networks
• Migrate to mesh when and where appropriate
• Dedicated 50msec restoration for those services
  requiring it (and willing to pay for it)
• Shared mesh restoration for resilient packet
  services (FR, ATM, IP)
• May save up to 60 in costs
• Send IP and Optical to marriage mediation
• Must learn to live together
• Divorce is not an option

13
General approach

• Virtualization
• Virtual has same functionality as a particular
  physical network, but does it through emulation
  (essentially software)
• Make physical networks more virtual
• To speed provisioning
• To allow faster upgrades
• Make virtual networks more physical
• To reduce overhead

14
Problem routers have limited visibility

• Routers do not naturally see
• Rings
• Connections
• Native IP is connectionless protocol
• Routers do see
• Ports and addresses (i.e., routing tables)
• Proprietary QoS queues

15
Brief historical background

• Early Internet was concerned only with mechanics
  of reliable data transfer
• Simple applications such as FTP, remote login
• Used software-based routers
• Later devices that could switch in hardware at
  levels 2 and 3 had to be deployed
• Layer 2 switching addressed bottlenecks in LANs
• Layer 3 switching addressed bottlenecks in layer
  3 routing by moving route lookup to high-speed
  hardware
• Issues
• Did not address service requirements for info in
  packets
• Based on shortest path only
• No consideration of jitter, delay, congestion
• Best effort utilizing algorithms in network
  components
• Little or no global control or optimization

16
The switching problem
OSI Reference Model

Application
Presentation
Session
Transport
Route/ Switch
Network
Knows about other workgroups
Router
Workgroup Switch
Hub
Data Link
Knows about local workgroup
Physical
Repeater
Doesnt know anything
17
The switching problem (continued)

• What does a switch do?
• Establishes a path through a network end-end
  (connection)
• Example circuit switch used in telephony
• No need for decisions at each point along the way

18
The switching problem (continued)

• What does a router do?
• Looks at incoming packet address and looks it up
  in table to find outgoing port
• No dedicated paths established (connectionless)
• Router does not know total path
• Dynamic paths
• Path for subsequent packets going to same
  destination may change due to congestion or other
  problems
• Requires seach
• Complexity O(log2 n), where n is number of
  entries in routing table

19
The switching problem (continued)

• IP traffic primarily routed
• ATM traffic primarily switched
• Permanent virtual circuit (PVC) fixed
• Switched virtual circuit (SVC) dynamic

20
The switching problem (continued)

• How to switch (route) packets with least
  expenditure of processing?
• How to allow different services to coexist on
  same IP network?
• At present, isochronous traffic (e.g., voice)
  does not work if network utilization greater than
  about 25
• Requires QOS (quality of service) or COS (class
  of service)
• How to allow different protocols on same network?
• IP
• ATM
• FR

21
The switching problem (continued)

• How to have a single packet forwarding method or
  paradigm while still allowing for different
  routing paradigms
• OSPF Open Shortest Path First
• PNNI Private Network to Node Interface or
  Private Network to Network Interface
• An ATM routing protocol

22
Desired solution elements

• Combine best of switching and routing
• Do routing once to find a path
• Record path elements
• Apply tag to subsequent packets with path
  information
• No need for looking into these packets to fetch
  addresses and do lookups at each router
• Complexity O(1), because indexing is used
• Initially called Tag switching or Label
  switching
• Similar (but not identical) to Post Office method
• Do handwriting recognition on a letter once
• Encode address info at bottom of envelope with
  bar code
• Use bar code to route letter through mail system

23

One of the many ways of getting from A to B

• BROADCAST Go everywhere, stop when you get to B,
  never ask for directions.
• HOP BY HOP ROUTING Continually ask whos closer
  to B go there, repeat stop when you get to B.
  Going to B? Youd better go to X, its on the
  way.
• SOURCE ROUTING Ask for a list (that you carry
  with you) of places to go that eventually lead
  you to B. Going to B? Go straight 5 blocks,
  take the next left, 6 more blocks and take a
  right at the lights.

Source Nortel
24
Label Switching

• Have a friend go to B ahead of you using one of
  the previous two techniques. At every road they
  reserve a lane just for you. At every
  intersection they post a big sign that says for a
  given lane which way to turn and what new lane to
  take.

LANE1
LANE2
Source Nortel
25
Basic idea behind label switching

• Set up virtual circuit between source and
  destination
• Assign numbers to each path element
• Copy numbers to packets
• Switch packet based on number
• Ingress router or host applies label
• Exit router strips it off

26
Basic idea behind label switching (continued)

• Forwarding of packets done using a short,
  fixed-length label rather than disassembly of
  complete address
• Addressing scheme different for different
  protocols (ATM, FR, IP, etc)
• Labels identify streams of traffic
• Label table much smaller than routing table
• Each label represents a set of destination
  addresses
• Packets with same label treated as a group, not
  individually
• Utilizes Time-To-Live (TTL) counter accurately
  maintained
• Idea is similar to PVCs and SVCs

27
Solution Multiprotocol Label Switching (MPLS)

• Layer 3 technology
• Works with any protocol, but primarily used for
  IP traffic
• Glues connectionless IP to connection-oriented
  networks
• IP to ATM
• IP to optical networks
• Referred to as shim layer
• Something between layer 2 and layer 3 to make
  them fit better

28
Solution (continued)

• Addresses problems of modern networks
• Speed
• Scalability
• Quality of Service (QoS) management
• Traffic engineering (TE)
• Multiprotocol

29
MPLS functions

• Mechanisms to manage traffic flows of various
  granularities
• Independent of layer 2 and layer 3 specs
• But serves as glue
• Maps IP addresses to fixed length labels to speed
  forwarding
• Interfaces to existing routing protocols such as
  OSPF
• Supports IP, FR, ATM layer 2 protocols

30
MPLS paths

• Utilizes label-switched paths (LSPs)
• Sequence of labels at every node from source to
  destination
• Each label represents a path between two nodes
• Set up in two ways
• Hop-by-hop
• Explicit routing
• Label establishment
• Prior to packet transmission (control-driven)
• Upon detection of a certain flow (data-driven)

31
MPLS devices

• LSR Label Switched Router
• High speed router (switch) in core of MPLS
  network
• Participates in establishment of LSPs
• LER Label Edge Router
• Operates at edge of access network and MPLS
  network
• Forwards traffic to MPLS network after
  establishing paths and attaching labels

32
Aggregating addresses in one label

• Aggregating addresses may be done in different
  ways
• Flow direction
• Traffic priority
• Traffic type
• Source address

Label Switched Path 225
Part of Label Information Base
Source Cellstream
33
There are many examples of label substitution
protocols already in existence

• ATM - label is called VPI/VCI and travels with
  cell.
• Frame Relay - label is called a DLCI and travels
  with frame.
• TDM - label is called a timeslot its implied,
  like a lane.
• X25 - a label is an LCN
• Proprietary PORS, TAG etc..
• One day perhaps Frequency substitution where
  label is a light frequency (or wavelength)?

34
Route at edge, switch in core

Source Nortel
35
Label creation methods

• Topology-based
• Uses normal processing of routing protocols
• Request-based
• Uses processing of request-based control traffic
• Traffic-based
• Uses reception of packet to trigger assignment
  and distribution of label

36
MPLS terminology

• Label short, fixed length, contiguous bits,
  locally significant (i.e., on a single link)
• Label switching router (LSR) Routers that use
  labels
• Traditional router
• ATM switch
• FR switch
• Optical switch
• Forwarding equivalence class (FEC) Same path and
  same treatment gt same label
• Label switched path (LSP) Particular path
  through network
• MPLS domain contiguous set of MPLS nodes in one
  administrative domain

37
MPLS terminology (continued)

• MPLS edge node ingress or egress node
• Label information base (LIB) label tables in
  each MPLS node which contain path information
  associated with labels
• Label distribution protocol (LDP) Method for
  distributing label information
• Flow flow of data from one application to
  another
• Stream Aggregate of one or more flows

38
Label switched path (vanilla)
39
Standard IP network
40
Normal routing of packet
41
Label distribution by MPLS
42
MPLS switching through network
43
Shim label for PPP traffic (most common in IP
networks)

• Packet structure

Link layer Header
SHIM
Network (IP) Layer Header
Payload
MPLS label (Mlabel)
Exper.
S
TTL
0
19
20
22
23
24
31
Exper.experimental COS
TTL time to live
S Bottom of stack (for multiple labels)
Source Cellstream
44
Labels can be stacked

Labels popped
225
Exper.
0
10
33
Exper.
0
7
105
Exper.
1
3
45
What happens when label looked up

• Next destination to which packet to be forwarded
  is found
• The correct operation required to be performed on
  packet before forwarding
• Replace top label stack entry with a new one
• Pop entry off stack (exposing next one down)
• Replace top label stack, push one or more new
  entries onto stack

46
Forwarding results of lookup

LSP 33
Label Switched Path 225
LSP 196
LSP 75
47
Labels can be merged

Label Switched Path 33
LSP 196
Label Switched Path 225
48
Labels can also be tunneled

LSP 33
LSP 33
LSP 99
LSP 225
LSP 225
49
Routing protocols in MPLS

• OSPF Open Shortest Path First
• Intended to yield better routing
• Based on link-state technology
• Allows Variable Length Subnet Masks (VLSM)
• Other enhancements
• BGP Border Gateway Protocol
• Purpose is to advertise to other routers what
  your network can route to (internally)
• IS-IS Intermediate System to Intermediate System
• Authentication between routers

50
Summary of motivations for MPLS

• Simplified forwarding based on exact match of
  fixed length label
• Initial drive for MPLS was based on existence of
  cheap, fast ATM switches
• Separation of routing and forwarding in IP
  networks
• Facilitates evolution of routing techniques by
  fixing the forwarding method
• New routing functionality can be deployed without
  changing the forwarding techniques of every
  router in the Internet
• Facilitates the integration of ATM and IP
• Allows carriers to leverage their large
  investment of ATM equipment

51
Summary of motivations for MPLS (continued)

• Enables the use of explicit routing/source
  routing in IP networks
• Can be easily used for such things as traffic
  management, QoS routing
• Promotes the partitioning of functionality within
  the network
• Move granular processing of packets to edge
  restrict core to packet forwarding
• Assists in maintaining scalability of IP
  protocols in large networks
• Improved routing scalability through stacking of
  labels
• Removes the need for full routing tables from
  interior routers in transit domain only routes
  to border routers are required
• Applicability to both cell and packet link-layers
• Can be deployed on both cell (eg. ATM) and packet
  (eg. FR, Ethernet) media
• Common management and techniques simplifies
  engineering

52
Generalized MPLS (sometimes referred to as MPlS)
or GMPLS

• MPlS Multiprotocol Lambda Switching
• Generalizes MPLS to deal with optical networking
• Photonic switches (PXCs)
• Optical Cross Connects (OXCs)
• Add/Drop Multiplexers (ADMs)
• DWDM
• Wavelength router
• Attempts to utilize as much of MPLS engineering
  as possible

53
GMPLS (continued)

• Requires rethinking of some concepts
• How label switching can be done
• What edge devices should see
• Solution Use control plane of MPLS
• Labels cant be applied to optical packets
• Must switch something labels can be applied to
  wavelengths
• To implement new functionality
• Dynamic provisioning (Point and click)
• Enhanced network survivability/restoration
• Flexible signaling and control architecture to
  support new applications

54
QoS and MPLS, MPlS

55
Current Inter-Networking Environment

• Current data Internet Protocol (IP) networks
  deliver packets on a best effort basis
• Meets requirements for data applications
• E-mail, file transfer, Web-browsing
• Does not meet requirements for real-time traffic
• Voice and video calls
• Collaborative conferencing
• Broadcast and multi-cast applications
• Provides no protection against cyberthreats such
  as Distributed Denial of Service (DDoS) attacks

56
Current Voice and Video Networks

• Voice networks
• Circuit-switched Time Division Multiplexed (TDM)
  networks, e.g., worldwide Public Switched
  Telephone Network (PSTN)
• Fixed connection bandwidth ( 64 Kbps), constant
  delay, no jitter, no data loss, highly available
• Video networks
• Predominantly based on Integrated Services
  Digital Network (ISDN)
• Connection-oriented with fixed bandwidth ( 64
  Kbps, 128 Kbps, 384 Kbps, 768 Kbps, 1.544 Mbps),
  constant delay, no jitter, no data loss, highly
  available
• Broadcast NTSC video distribution
• 45 Mbps T3-based TDM network

20-year-old technology, deployed in the mid-1980s
57
Enhancing Internet Protocol (IP) Networks To
Support A Variety Of Applications
58
Challenge Enhancement of IP Infrastructure to
Support Diverse Set of Applications

• Service providers and network managers operating
  multiple networks to support range of
  applications
• This is not desirable from economic and
  maintenance standpoint
• IP infrastructure devices becoming cheaper due to
  proliferation of the public Internet and private
  networks
• Routers/switches and transmission
• Current IP infrastructure needs enhancement to
  support voice, video, and data at acceptable
  levels
• Flow of real-time bit streams

This is the challenge for the decade
59
Real-Time / Multimedia Requirements

• Support for a range of diverse applications
• Support for a range of bandwidth
• E.g., 128 Kbps collaborative video
  conferencing to 45 Mbps video-on-
  demand
• Support for a range of performance for voice,
  video, multimedia, critical data
• Delay, delay variation, packet loss
• Support a range of communication models
• Point-to-point, multipoint, multicast, broadcast
• Use of QoS for cybersecurity looks promising

60
Solution Alternatives

• Massive overbuild
• Brute force approach
• Feasible in good old POTS days
• Due to fractal nature of Internet traffic,
  difficult to know how much capacity is enough
• Fractal self-similar on multiple time scales
• Quality of Service (QoS) / Class of Service (CoS)
• Preferentially routes packets based on type of
  traffic they carry
• Does require software and / or hardware upgrades
• Complex nature of Internet and other networks
  makes prediction of performance difficult

61
Fractal Nature of Internet Traffic
Packets/100 msec

Packets/1 sec
Packets/10 sec
Packets/60 sec
Source Willinger and Paxson, 1998
62
Internet Time Scales

Fractals Long-Range Dependency
Multifractals Effects of Network Transport
Protocols
Diurnal and Other Effects
1 ms
10
100
1 s
10
100
1,000
104
Measurement Time
63
Invariants in Data Traffic
64
Determinants of Traffic Statistics

• Application structure
• User behavior
• File sizes

Monofractal scaling at time scales gt 300 msec
?
WANs and LANs

• Network control
• mechanisms

Multifractal scaling at time scales lt 300 msec
WANs only
?
65
Different Protocols Mean Different Time Scales
Multiple packet streams
Minutes, hours

http
ftp
smtp
. . .
Traffic granularity
Time scale
Transmission Control Protocol (TCP)
Packet streams
100s ms
Internet Protocol (IP)
Packets
ms
Frames, bits
Ethernet
100s ns
66
Quality of Service (QoS)As A Solution
67
What is Class of Service / Quality of Service ?

• CoS
• Classification of packets for the purpose of
  treating certain classes or flows of packets in a
  particular way compared to other packets

• QoS
• QoS defined as users experience over a network
  connection

Clearly, QoS will require some type of CoS
68
QoS Metrics

• Network delay Also known as latency
• Delay variation Also called Jitter
• Throughput Packet rate (average, peak)
• Packet loss rate Maximum rate at which packets
  can be discarded
• Network service availability

69
QoS / CoS Approach

• Develop new protocols to support real-time
  applications
• Split problem into access, backbone
• Develop appropriate access, backbone QoS
• Map access QoS (classes) into backbone QoS
  (classes)
• Resolve issues to assure smooth end-to-end QoS as
  seen by user

70
Real-Time Application Protocols
71
New Protocols Providing Real-Time Support for IP
Networks

• New protocols developed for routing and switching
  of real-time traffic
• Multi-Protocol Label Switching (MPLS)
• New protocols to support transport of real-time
  traffic
• Real-Time Transport Protocol (RTP)
• Real-Time Control Protocol (RTCP)
• Real-Time Streaming Protocol (RTSP)
• New protocols to support real-time applications
• H.323 and Session Initiation Protocol (SIP)

72
Real-Time Applications Protocol Stack
Presentation
G.729(A)/G.723(.1)G.711
Session
H.323/SIP/MGCP/RSVP/RTSP
Transport
RTP-RTCP/UDP
Network
Network
Link
IP (Use of IP Header for DiffServ)
Physical
- - - - - -
73
MPLS for Real-Time Traffic

• Switching technology to support real-time flows
  in IP networks
• Designed to perform similar function to ATM
  Virtual Circuits
• Label Switched Path (LSP) pre-established to
  support specific QoS
• Label Distribution Protocol (LDP) used to
  accomplish this

74
Stages of MPLS processing

• Customer premises router supplies QoS info with
  each packet
• Packet header examined at the entry point to MPLS
  network
• A label created by the edge router indicating
  packet classification
• Core routers perform switching based on labels
• Only labels examined at intermediate points to
  support high-speed switching
• Less work involved compared to full packet
  processing

75
MPLS for Real-Time Traffic (Concluded)

• IP VPN (Virtual Private Network)
• A second unique label used to identify specific
  VPN packets
• Works because label lookup is much faster than
  full address decoding
• Limitation is that number of labels ltlt number of
  Internet addresses

76
End-to-End QoS Model
Access Network
Backbone
Access Network
Applications
Applications
Presentation
Presentation
Internet Protocol (IP) or Asynchronous Transfer
Mode (ATM)
Session
Session
Transport
Transport
Network
Network
Data Link
Data Link
Physical
Physical
ATM QoS or IP QoS Differentiated Services
(DiffServ)/ MPLS
802 Subnet Bandwidth Management
(SBM) ReSerVation Protocol (RSVP)
802 Subnet Bandwidth Management
(SBM) ReSerVation Protocol (RSVP)
77
End-to-End QoS Model (Concluded)

• Access QoS
• Must be granular enough to differentiate service
  requirements of multiple traffic streams
• Bandwidth control and traffic policing required
  at network entry points
• Backbone QoS
• Backbone must provide enough transport and
  control to satisfy the service levels promised to
  customers
• IP QoS works on aggregate flows of traffic
• ATM QoS works on specific flows

78
Two Locations for QoSAccess and Backbone
79
Access QoS

• Access networks
• Customer premises networks
• Predominantly Ethernet LANs with IP
• Shared/switched Ethernet to desk-top
• Fast/Gigabit Ethernet backbone
• No industry consensus on how to manage CoS/QoS at
  this level
• Some efforts made
• Signaling between client and bandwidth manager
  (RSVP)
• Priority of frames at Ethernet level (802.1p) to
  support QoS

80
Backbone QoS Two Options

• ATM QoS
• Well-defined QoS for ATM service
  (connection-oriented)
• IP QoS
• In evolutionary stage
• A range of protocols and architecture developed
  to support IP QoS
• Primary mechanisms within the switches/routers
  used are
• Queuing of traffic based on classes
• Different forwarding priorities
• Different discard priorities

81
Backbone QoS ATM Wide Area Network (WAN)

• Each ATM connection established to meet a
  specific QoS requirement
• QoS specified during connections set-up time and
  can be re-negotiated during a connection
• QoS in ATM networks characterized by a set of
  parameters
• Max Cell Transfer Delay (CTD)
• Cell Delay Variation (CDV)
• Cell Loss Ratio (CLR)
• Cell Error Ratio (CER)

82
Backbone QoS ATM Wide Area Network (WAN)
(Concluded)

• A range of QoS-based services
• Constant Bit Rate (CBR)
• Variable Bit Rate real-time (VBRrt)
• Variable Bit Rate non-real-time (VBRrt)
• Available Bit Rate (ABR)
• Unspecified Bit Rate (UBR)

83
DiffServ and QoS
84
DiffServ Model

• Problem how do we know what classes of service
  are needed in order for user to experience
  desired QoS?
• DiffServ model tries to answer this
• Defines an architecture for a set of service
  classes and QoS mechanisms for packet handling in
  those classes
• Not the same thing as MPLS
• Service providers providing Class of Service at
  ingress and egress points of MPLS IP networks
  trying to conform to DiffServ QOS

85
DiffServ Model (Concluded)

• Provides a simple and coarse method of
  classifying services of various applications
• Type of Service (ToS) field in IP version 4 has
  been renamed as DS (Differentiated Services)
  field (6 bits used)
• Following types of classes supported
• Expedited Flows (EF)
• Assured Forwarding (AF) Class
• Network edge devices assign DiffServ bits to
  packets for consistent treatment within the
  network
• Transit routers and switches will usually
  separate the traffic based on DiffServ bits into
  queues

86
Classes of Services in IP Networks

• Generally four traffic classes need to be
  supported at entry/exit points in IP networks
• Expedited flow For voice and network control
• Real-time traffic Mostly video applications
• Critical data Mission-critical data
  applications
• Best effort E-mail and browsing

87
Current IP CoS/QoS Approaches for Backbone

• Three basic approaches by service providers in
  near term
• No CoS/QoS support?pure IP routed backbone with
  Gigabit routers/Synchronous Optical Network
  (SONET) Transmission
• Support DiffServ-compliant CoS/QoS at
  Ingress/Egress points with no CoS/QoS support in
  the core MPLS backbone
• Support DiffServ-compliant CoS/QoS at
  Ingress/Egress points and use ATM-based QoS in
  the networking backbone
• Future IP-based QoS in backbone

88
Option 1 No QoS Support in Backbone

• Variant of massive overbuild strategy
• Private networks only
• MPLS
• Gigabit routers
• SONET
• High-speed (OC48)
• Ensures low jitter, low utilization

89
Option 2 DiffServ Compliant / No CoS/QoS
Support in Backbone

• Also for private networks
• IP QoS supported only at entry and exit points of
  MPLS networks
• Entry and exit points represent bottlenecks, and,
  therefore, need priority management
• Very little traffic congestion in the backbone
  Gigabit routers / Gigabit Dense Wavelength
  Division Multiplexing (DWDM) pipes
• May use Packet-over-SONET (POS)
• Typically 50 msec delay coast-to-coast

90
Option 3 DiffServ Compliant CoS/QoS at
Ingress/Egress Points / ATM-Based QOS

• IP service provided over ATM cloud
• ATM switches upgraded to support MPLS
• ATM services utilized to obtain desired QoS
• SONET interfaces
• Transit delays of 70 msec in backbone
  coast-to-coast

91
Future All-IP Networks With IP Over Optical

• Likely goal will be IP over DWDM, bypassing ATM
  and SONET
• QoS will have to be functional in this environment

Internet Protocol
Encapsulation
H.323/SIP/MGCP/RSVP/RTSP
PPP/HDLC
SRP
1/10 GE-MAC
ATM
SDL
Optical Interface
SONET/SDH SDL-PHY
H.323/SIP/MGCP/RSVP/RTSP
SONET/SDH
SONET/SDH
1/10 GE-PHY
ATM-PHY
WDM / DWDM
Packet over SONET (PoS) PPP does L2 Functions
Dynamic Packet Transport (DPT) Spatial
Reuse Protocol (SRP) Intended for Ring
Architecture
Gigabit Ethernet (GE)
Asynchronous Transfer Mode (ATM)
Simple Data Link (SDL)
Source Cisco/Tomsu Schmutzer
92
Work To Be Done

• IP QoS implementation still evolving
• No industry consensus on how IP LANs and IP MPLS
  WANs will work together to offer end-to-end QoS
• Number of traffic flows/priorities to be
  supported at entry/exit points
• Admission control and traffic management at
  entry/exit points of backbone need to be
  carefully managed
• Role and value of MPLS support for CoS/QoS in the
  core switches/routers not clear
• Need for QoS support from MPLS?
• Will depend on architecture
• IP over DWDM?

93
Cyber Security and QoS
94
Mitretek Laboratory Work on QoS and Cyber Security

• Cybersecurity has become issue of great
  importance for Government and private sector
• Mitretek has developed extensive capabilities to
  study network performance under QoS
• Laboratory
• Analytic / simulation
• Capabilities can also be used to study various
  cyber attacks and performance of IP networks
  under congestion conditions
• DDoS attacks
• Congestion resulting from damage to links,
  switches, routers

95
QoS and Cyber Attack Modules
96
Mitretek Lab Work on QoS and Cyber Security

• Three-node test to show effect of QoS on network
  flooding by DDoS attack

97
Link Utilization Near 100 Percent
98
Results of QoS
Video Without QoS
Video with QoS
99
Analytical Studies of Networks Under Congestion
and Cyberattack

• Questions of interest in todays environment
• How vulnerable are large networks to attack?
• Can we predict the performance of a network under
  attack?
• Mitretek has developed an analytic model called
  the IP Network Performance and Analysis Tool
  (IP-NPAT) and an OPNET simulation model to
  address these types of questions
• Analyzes IP networks under variety of conditions
• Cyber attacks
• Implementation of new programs or protocols
• Developed to support Government agencies

100
Analytical Studies of Networks Under Congestion
and Cyberattack (continued)

• Analytic techniques allow Mitretek to study
  network congestion in the presence of
  heavy-tailed traffic distributions
• Waiting time CDF for links cannot be calculated
  using queuing theory when traffic distributions
  are heavy-tailed
• Mitretek has developed a technique called the
  Transform Approximation Method (TAM) and its
  associated numerical procedure, called the TAM
  Recursion Method
• Allows end-to-end waiting times to be estimated
  in congested networks

101
Analytical Studies of Networks Under Congestion
and Cyberattack (Concluded)

• Used in conjunction with laboratory studies
• Comparison with simulations has verified accuracy
  of analytic methodology and tools

102
Comparison of Analytic and Simulation Results
103
Future enhancements/applications

• Analytic model expanded to include
• DiffServe
• Voice, Video, Data packets
• MPLS
• Used to design secure networks