Submitted to the Department of Computer Science and Engineering of Jagannath University in partial fulfillment of

the requirements for the degree of M.Sc. in CSE

i
Abstract

ii
The world has emerged as a global village. Internet has brought an amazing change in the era of
modern communications. Demand for multimedia applications and an ever-increasing amount
of VoIP traffic have increased data rate and bandwidth requirements. It has become a big
challenge to provide best quality applications.

Traffic engineers are working hard over Internet Protocol (IP) and Routing Protocols (RPs) in
order to cope with this challenge. IP networks have offered these services efficiently until now,
but there are several issues with IP routing that affect the Quality of Service (QoS). Multi-
Protocol Label Switching (MPLS) has emerged as an agile technology which promises effective
security together with high speed data delivery.

MPLS networks use switching instead of conventional routing. MPLS doesn’t replace existing IP
networks, but guarantees instead better QoS in existing IP networks as well as future routing
technologies, providing thus a suitable environment for Traffic Engineering (TE). The Class of
Service (CoS) architecture provided by MPLS can easily be coordinated with IP QoS
mechanisms.

Traffic can be switched based on resource utilization and network performance instead of using
static routing techniques such as selecting a path with least cost, optimizing thus the MPLS
network for heavy application at desired quality.

Keywords: IP, MPLS, TE and QoS

Acknowledgements

iii
At first remember the Almighty Allah, since he gave us the strength to complete the project. I
would like to thanks our supervisor, Professor Dr. Uzzal Kumar Acharjee, Director, Department
of Computer Science and Engineering, Jagannath University. Without his support and guidance,
this project could not elevate up this level of development from my point of view.

And I would like to thanks all the Faculty members, Department of Computer Science and
Engineering, Jagannath University for their valuable time spend.

We would like to express our sincere and cordial gratitude to the people those who have
supported us directly, purveyed mental encouragement, evaluated and criticized our work in
several phases during the development of this internship and for preparing this paper
indirectly.

iv
 Table of Contents

List of Figures.................................................................................................................................vii
List of Tables..................................................................................................................................viii
List of Abbreviations........................................................................................................................ix
Chapter-1 Introduction....................................................................................................................1
1 Introduction..................................................................................................................................2
1.1 Project Aim.............................................................................................................................2
1.2 Project Objectives...................................................................................................................2
1.3 Motivation..............................................................................................................................2
1.4 Project Structure....................................................................................................................3
1.5 Project Outcome....................................................................................................................3
Chapter-2 IP Routing........................................................................................................................4
2 Background...................................................................................................................................5
2.1 Type of Service.......................................................................................................................5
2.1.1 Connection Oriented Service...........................................................................................5
2.1.2 Connectionless Service....................................................................................................5
2.2 Types of Transmission............................................................................................................5
2.2.1 Routing.............................................................................................................................6
2.2.2 Switching..........................................................................................................................6
2.3 Types of Devices.....................................................................................................................6
2.3.1 Router...............................................................................................................................6
2.3.2 Switch/Hub/Bridge...........................................................................................................6
2.3.3 Server/Workstation..........................................................................................................6
2.3.4 Links..................................................................................................................................6
2.4 Types of Networks..................................................................................................................6
2.4.1 Local Area Network..........................................................................................................7
2.4.2 Wide Area Network..........................................................................................................7
2.4.3 Metropolitan Area Network............................................................................................7
2.5 Types of Protocols..................................................................................................................7
2.5.1 Routing Protocols.............................................................................................................7
2.5.2 Routed Protocols..............................................................................................................7
2.5.3 Messaging Protocols........................................................................................................7
2.6 Internet Protocol....................................................................................................................7
2.6.1 IP Operations....................................................................................................................8
2.6.2 IP Encapsulation...............................................................................................................8
2.6.3 IP Header..........................................................................................................................8
2.7 Routing Mechanism................................................................................................................9
2.8 Autonomous System..............................................................................................................9
2.8.1 Stub AS.............................................................................................................................9
2.8.2 Transit AS.........................................................................................................................9
2.8.3 Multi-Homed AS.............................................................................................................10

v
 2.9 Routing Techniques..............................................................................................................10
2.9.1 Static Routing.................................................................................................................10
2.9.2 Distance-Vector Routing................................................................................................10
2.9.3 Link State Routing..........................................................................................................10
2.9.4 Hybridized Routing.........................................................................................................11
2.10 IP Network’s Limitations....................................................................................................11
2.10.1 Scalability.....................................................................................................................11
2.10.2 Class of Service.............................................................................................................11
2.10.3 Lack of Route Recovery................................................................................................12
2.10.4 Lack of QoS...................................................................................................................12
2.10.5 Inaccessibility of Bandwidth........................................................................................12
Chapter-3 Problem Statement......................................................................................................13
3.1 Important Issues...................................................................................................................14
3.2 Project Goal..........................................................................................................................15
Chapter-4 Multi-Protocol Label Switching....................................................................................16
4.1 Background...........................................................................................................................17
4.2 MPLS Concepts.....................................................................................................................17
4.3 MPLS Applications................................................................................................................17
4.3.1 Traffic Engineering.........................................................................................................17
4.3.2 Path Management.........................................................................................................18
4.3.3 Virtual Private Network.................................................................................................18
4.3.4 Multi-Protocol Support..................................................................................................19
4.4 MPLS Header........................................................................................................................19
4.5 MPLS Operations..................................................................................................................19
4.6 MPLS Network Architecture.................................................................................................20
4.6.1 Forwarding Component.................................................................................................20
4.6.2 Control Component.......................................................................................................20
4.6.3 Control Plane..................................................................................................................21
4.6.4 Data Plane......................................................................................................................21
4.6.5 Label Switching Router..................................................................................................21
4.6.6 Label Edge Router..........................................................................................................21
4.6.7 Label Switching Path......................................................................................................21
4.6.8 Label Distribution Protocol............................................................................................22
4.6.9 MPLS Label.....................................................................................................................22
4.6.10 Forwarding Equivalence Class......................................................................................23
4.7 MPLS Operation Modes.......................................................................................................23
4.7.1 Frame Mode...................................................................................................................23
4.7.2 Cell Mode.......................................................................................................................25
4.8 Loop Detection and Prevention........................................................................................27
4.8.1 Frame Mode...................................................................................................................28
4.8.2 Frame Mode: Control Plane Loop Prevention...................................................................28
4.8.3 Data Plane Mode............................................................................................................28
4.9 LSR Mode..............................................................................................................................29
4.9.1 Label Distribution Mode................................................................................................29

vi
 4.9.2 Label Retention Mode....................................................................................................29
4.9.3 Control Mode.................................................................................................................29
4.10 MPLS Packet Forwarding....................................................................................................30
Chapter-5 MPLS Traffic Engineering..............................................................................................31
5.1 Overview...............................................................................................................................32
5.1.1 Basics of TE.....................................................................................................................32
5.2 MPLS-TE................................................................................................................................33
5.2.1 Overview........................................................................................................................33
5.2.2 Load Balancing...............................................................................................................33
5.2.3 Link Congestion..............................................................................................................34
5.2.4 Link Protection...............................................................................................................35
5.3 Advantages of MPLS-TE........................................................................................................36
5.4 MPLS-TE Operations.............................................................................................................36
5.4.1 Distribution of Link Information....................................................................................36
5.4.2 Computation of Path......................................................................................................36
5.4.3 LSP Signaling...................................................................................................................37
5.5 MPLS-TE Classes...................................................................................................................37
5.5.1 Traffic Oriented..............................................................................................................37
5.5.2 Resource Oriented.........................................................................................................37
5.6 RSVP and TE..........................................................................................................................37
5.6.1 RSVP Packet Format.......................................................................................................38
5.6.2 RSVP Object Fields.........................................................................................................38
5.7 RSVP Message Types............................................................................................................39
5.7.1 RSVP Path Message........................................................................................................39
5.7.2 RSVP Path Tear Message................................................................................................39
5.7.3 RSVP Path Error Message...............................................................................................39
5.7.4 RSVP Reservation Request Message..............................................................................40
5.7.5 RSVP Reservation Confirm Message..............................................................................40
5.7.6 RSVP Reservation Tear Message....................................................................................40
5.7.7 RSVP Reservation Error Message...................................................................................40
5.8 RSVP Reservation.................................................................................................................41
5.8.1 Sender Selection............................................................................................................41
5.8.2 Session Reservation.......................................................................................................41
5.8.3 Reservation Styles..........................................................................................................41
5.9 Quality of Service..................................................................................................................42
5.10 Internet Service Models.....................................................................................................42
5.10.1 Best-Effort Service Model............................................................................................42
5.10.2 Integrated Service Model.............................................................................................43
5.10.3 Differentiated Service Model.......................................................................................44
5.10.4 Hybrid Model...............................................................................................................44
Chapter-6 Label Management.......................................................................................................45
6.1 Label Distribution Protocol...................................................................................................46
6.1.1 LDP Packet Format.........................................................................................................46
6.1.2 LDP Message Format.....................................................................................................47

vii
 6.1.3 LDP Type-Length-Value Format.....................................................................................47
6.2 LDP Messages.......................................................................................................................47
6.2.1 Notification Message.....................................................................................................48
6.2.2 Hello Message................................................................................................................48
6.2.3 Initialization Message....................................................................................................48
6.2.4 KeepAlive Message........................................................................................................48
6.2.5 Address Message...........................................................................................................48
6.2.6 Address Withdraw Message..........................................................................................48
6.2.7 Label Mapping Message................................................................................................48
6.2.8 Label Request Message..................................................................................................48
6.2.9 Label Abort Request Message.......................................................................................48
6.2.10 Label Withdraw Message.............................................................................................49
6.2.11 Label Release Message................................................................................................49
6.3 Merging Labels.....................................................................................................................49
6.4 Label Distribution Modes.....................................................................................................49
6.4.1 Independent Label Distribution Control........................................................................50
6.4.2 Ordered Label Distribution Control...............................................................................50
6.5 Label Retention....................................................................................................................50
6.6 Free Labels............................................................................................................................51
6.7 Label Space...........................................................................................................................51
6.7.1 Per-Interface Label Space..............................................................................................52
6.7.2 Per-Platform Label Space...............................................................................................52
6.8 Label Stacking.......................................................................................................................52
Chapter-7 Conclusions...................................................................................................................54
7.1 Issues....................................................................................................................................55
7.1.1 Path Capacity and Multi-Class Traffic............................................................................55
7.1.2 Load Dependent Parameters and Path Capacity...........................................................55
7.1.3 Load Dependent Parameters and Non-Linear Models..................................................55
7.2 Project Conclusions..............................................................................................................56
7.3 Future Work.........................................................................................................................56
References.....................................................................................................................................57

viii
 List of Figures
Figure 2. 1 IP Datagram...................................................................................................................................................8

Figure 4. 1 Traffic Volumes in Conventional IP Network..............................................................................................18

Figure 4. 2 MPLS Shim Header......................................................................................................................................19
Figure 4. 3 MPLS Operations.........................................................................................................................................20
Figure 4. 4 LSP through a MPLs Network......................................................................................................................22
Figure 4. 5 MPLS Label Format.....................................................................................................................................23
Figure 4. 6 Frame Mode MPLS Label Assignment and Distribution.............................................................................24
Figure 4. 7 Frame Mode MPLS Forwarding..................................................................................................................25
Figure 4. 8 Cell Mode MPLS Label Imposition..............................................................................................................26
Figure 4. 9 Cell Mode MPLS Domain.............................................................................................................................26
Figure 4. 10 Cell Mode MPLS Label Allocation and Distribution..................................................................................27
Figure 4. 11 Loop Detection Processes......................................................................................................29

Figure 5. 1 IP Forwarding Networks..............................................................................................................................32

Figure 5. 2 Shortest Path Computations.......................................................................................................................34
Figure 5. 3 Over-Utilized Links......................................................................................................................................35
Figure 5. 4 Primary Path Failures..................................................................................................................................35
Figure 5. 5 Path Message and Path Error Message......................................................................................................40
Figure 5. 6 Reservation Request Message and Reservation Error Message..............................................41

Figure 6. 1 Downstream on Demand Label Advertisements........................................................................................49

Figure 6. 2 Unsolicited Downstream Label Advertisements.........................................................................................50
Figure 6. 3 Liberal Label Retention Modes...................................................................................................................51
Figure 6. 4 Push and Pop Operations............................................................................................................................53

ix
 List of Tables
Table 2. 1 IP Header.....................................................................................................................................8

Table 5. 1 RSVP Packet Format.....................................................................................................................................38

Table 5. 2 RSVP Message Types....................................................................................................................................38
Table 5. 3 RSVP Object Fields....................................................................................................................39

Table 6. 1 LDP Packet Format.......................................................................................................................................46

Table 6. 2 LDP Message Format....................................................................................................................................47
Table 6. 3 LDP Type-Length-Value Format...................................................................................................................47

x
 List of Abbreviations
Acronym Abbreviation

ARP Automatic Repeat Request

ATM Asynchronous Transfer Mode

BGP Border Gateway Protocol

BoS Bottom of Stack

CBR Constant Bit Rate

CEF Cisco Express Forwarding

CoS Class of Service

CRC Cyclic Redundancy Check

DV Distance Vector

DBMS Data Base Management System

EGP Exterior Gateway Protocol

EIGRP Enhanced Interior Gateway Routing Protocol

FEC Forward Error Correction

FTP File Transfer Protocol

GWT Google Wireless Transponder

ISP Internet Service Provider

IP Internet Protocol

IOS Internet Operating System

IGP Interior Gateway Protocol

ISIS Intermediate System to Intermediate System
IPTV Internet Protocol Television

xi
IPV4 Internet Protocol Version 4

ITU International Telecommunication Unit

IHL Internet Header Length
IETF Internet Engineering Task Force
IBM International Business Machines
ISDN Integrated Service Digital Network
LAN Local Area Network
LDP Label Distribution Protocol
LSR Label Switch Router
LSP Label Switch Path
LFIB Label Forwarding Information Base
MPLS Multiprotocol Label Switching
MAN Metropolitan Area Network
MTU Maximum Transmission Unit
MAC Medium Access Control
MIB Management Information Base
NMS Network Monitoring System
NAT Network Address Translation
NOC Network Operation Centre
OSI Open System Interconnection
OSPF Open Shortest Path First
OAM Object Access Method
PDU Packet Distribution Unit
PVC Permanent Virtual Circuit
PHB Per Hop Behavior
POP Point of Presence
PSTN

xii
 Public Switch Telephone Network
PPP
Point to Point Protocol
PDU
Packet Distribution Unit

QoS Quality of Service

RSVP Resource Reservation Control Protocol

RIP Routing Information Protocol

SMS Short Message Service

SNMP Simple Network Management Protocol

SLA Service Level Agreement

TE Traffic Engineering

TCP Transmission Control Protocol

TTL Time To Live

LIB Label Information Base

TDM Time Division Multiplexing

TDP Tag Distribution Protocol

ToS Type of Service

UDP User Datagram Protocol

VLAN Virtual Local Area Network

VoIP Voice over Internet Protocol

VPN Virtual Private Network

xiii
Chapter-1 Introduction

1
1 Introduction
Internet has grown tremendously over the years and this growth has also raised concerns for
good management and performance of the packet switched network communications.
Engineers are developing new techniques, architectures and protocols to best support these
new means of communication. Alongside this development, a continuous effort is being put on
planning future protocols or changing existing ones.

Many fascinating devices and applications have emerged with the boom of Internet. To live up
to users’ expectations, application efficiency and performance needs to be consistent. IP
networks are serving well existing user application and their demands, but there are some
elements that often cause problems for an ever-growing range of applications.

All issues relating to communication are resolved through Traffic Engineering (TE). TE is a set of
tools and techniques for managing networks data comprehensively. TE helps a lot in serving
users and their expectations. Traffic engineers working over the years have come up with a
technology known as Multi-Protocol Label Switching (MPLS). MPLS uses both new techniques
and existing IP routing techniques in order to improve performance of data communications.

1.1 Project Aim

Main aim of the project is to discuss TE in accordance with improving Quality of Service (QoS)
and to show how MPLS is best suited for resolving existing QoS issues.

1.2 Project Objectives

Following objectives will help achieve the aim of the project.

• Identify the current state of communication techniques through TE.

• Identify the IP network’s limitations with influence on TE.
• Identify the features in MPLS which promise improved QoS.
• Investigate the mechanism used by MPLS.

1.3 Motivation
IP networks have been supporting several great applications for many years, but MPLS emerged
as a new technology that improves overall QoS. Many problems of IP networks are promised to
be solved and improved by the MPLS technology.

Many new protocols and techniques will emerge on this new communication era that supports
VPN tunneling, TE, multicasting and many more. Stock holders, i.e., Internet Service Providers
(ISPs) and enterprises with large autonomous networks need to implement MPLS in order to
continue competing in a challenging market and for efficiently serving their clients.

Development of real-time applications by leading market vendors has triggered a need for an
efficient technology which will fulfill the ever-increasing user demands. Real-time applications
like IPTV need extensive bandwidth and moreover, a quality network. Therefore, efficient

2
solutions like efficient bandwidth utilization provided by MPLS are the answer to uncertainty
and deliver a promise for quality.

1.4 Project Structure

Overall structure of the project report is as follows:

Chapter 1 contains information about the project objectives, Motivation and its structure
Chapter 2 discusses IP networks. It provides a detailed overview of the routing mechanism and
different protocols used in IP routing. Chapter 3 highlights the main problems for which this
research is conducted. Chapter 4 explains the MPLS technique, its components, functions and
protocols. Chapter 5 discusses aspects involved in TE that enhance QoS in MPLS. Chapter 6
gives a comprehensive study of label management mechanism, i.e. how labels are added,
removed and managed. Finally, Chapter 7 concludes this re-search work by highlighting some of
the MPLS issues, presenting our conclusions and thoughts for future work.

1.5 Project Outcome

MPLS networks’ performance to serve interested readers and researchers in terms of different
routing protocols.

IP networks’ limitations will be identified and features of MPLS which improve network
performance are elaborated in detail. Finally, a picture will be drawn to conclusively support
that MPLS is far better than IP networks in terms of efficiency, performance and quality of
network transmission. It will also help improve the theoretical background for future industrial
projects in MPLS.

3
Chapter-2 IP Routing

4
2 Background
Computer systems were “standalone” at times and data was manually shared between them. In
turn, this prompted for a need of inter-networking between these systems in order to ensure
automatic and easy data exchange. Emergence of huge networks has compelled computers to
communicate through various protocols. A protocol is said to form a set of rules that govern
data communication.

These protocols provide services to either a wired or a wireless medium over a physical or
logical channel. Several techniques and devices have been developed to support efficient
networking services.

A large range of applications have been developed that efficiently share information and pro-
vide services both to officials (business-to-business, resource sharing, data storage, print and
scan utilities, conferencing, etc.) and to home users (fun and multimedia, live video streaming,
SMS and interactive chat services). These applications work regardless of media and location,
i.e., the wireless and wired medium supports these services and even mobile devices that
continuously change their location.

2.1 Type of Service

Data communication is supported by two different techniques depending on their needs, each
with its own advantages and disadvantages.

2.1.1 Connection Oriented Service

It is a service that requires a communication session to be established before the data can be
sent such as communication through Public Switched Telephone Networks (PSTNs). The
connection-oriented service provides a reliable communication mechanism. However, unlike
PSTN’s physical dedicated line, virtual links are used to define a connection in IP networks. Data
is sent as a stream and this service ensures that data arrives in the same order as it was sent.

2.1.2 Connectionless Service

It is a service that doesn’t require an established session prior to data communication. Sender
starts sending data (datagrams) regardless of its delivery status. It is the receiver’s responsibility
to rearrange it once it has reached its destination. It doesn’t provide reliability; however, it is
considered a faster mean of communication.

2.2 Types of Transmission

Data is transmitted using the following two techniques depending on the architecture of the
network.

5
2.2.1 Routing
It is a process of finding the path and move data from a source to a destination in computer
networks. It is performed in two or several networks, e.g. telephone network, electronic data
networks (the Internet) and transportation networks

2.2.2 Switching
It is a process of moving packets between devices within the same network. Sender and
receiver are physically connected via a central device which performs switching. This device
must know the physical address of the devices; this address is also known as the Media Access
Control (MAC) address

2.3 Types of Devices

A network is set of devices inter-connected with links The following device types make up a
network.

2.3.1 Router
A router is a device which transmits data over two or more networks. It uses destination ad-
dresses and some special service information to forward data to a specific path. It directly or
indirectly connects to devices belonging to different networks. In general terms, the router is
the device that performs the routing

2.3.2 Switch/Hub/Bridge
It is a category of devices which transmit data between two or more directly connected devices.
The switch/hub/bridge needs to know the physical address of the connected devices. Switches
use a communication technique known as switching

2.3.3 Server/Workstation
A server is a computing device that provides services to other computers. The computers using
services from the server are known as workstations or clients.

2.3.4 Links
Links are the actual physical connections between internetworking devices. These links may be
wired (guided, i.e., Twisted-Pair Cable, Coaxial Cable, Fiber-Optic Cable) or wireless (unguided,
i.e., Radio Waves, Microwaves, Infrared)

6
2.4 Types of Networks
Depending on the territory and physical layout of a network, a network can be one of the fol-
lowing:

2.4.1 Local Area Network

Local Area Network (LAN) is a group of computers and associated or connected devices that
share resources of servers through a single wired or wireless link within a small geographical
area such as an office building, usually owned by a single authority the technology used in LAN
is mostly Ethernet.

2.4.2 Wide Area Network

Wide Area Network (WAN) is a network that covers a larger geographical area and is often used to
connect two or more LANs or remote computers. Routers are the inter-connecting device used in WANs.
WANs may be privately owned or rented but are usually a sharing of public networks

2.4.3 Metropolitan Area Network

Metropolitan Area Network (MAN) is a network that interconnects computes situated in a geographical
area larger than area covered by a LAN but smaller than the area covered by a WAN. It is usually used to
interconnect networks in a small town

2.5 Types of Protocols

Communication protocols are generally categorized as follows:

2.5.1 Routing Protocols

These are software protocols that enable a router to dynamically advertise routing information to other
routers and learn their routing information, determine available routes and finally determining the most
suitable route to the destination. Known routing protocols are Routing Information Protocol (RIP), Open
Shortest Path First (OSPF), Interior Gateway Routing Protocol (IGRP) and Border Gateway Protocol (BGP)

2.5.2 Routed Protocols

Routed protocols are generally used as reference or address to data itself, like Internet Protocol (IP).
They represent the data which is being transported. Some of the routed protocols are SNMP, SMTP and
Telnet

2.5.3 Messaging Protocols

7
These are the type of protocols that perform error reporting. If a host or server is unreachable or a ser-
vice is unavailable these protocols are used by the operating system of the communicating device to
inform the other connected devices about the situation. These are Internet Control Message Protocol
(ICMP, ICMPV6) and Internet Group Management Protocol (IGMP)

2.6 Internet Protocol

Internet Protocol (IP) is a routed protocol and is used to transmit data across a network using the
Transmission Control Protocol (TCP)/IP protocol Suite. It treats data as small units called datagrams and
takes these datagrams from a source to a destination (host-to-host) with the help of their IP addresses.
To transmit datagrams, IP defines an addressing mechanism and a datagram encapsulation structure

IP has two versions; however, we will only discuss version 4, which is most dominant of the two.

• IPv4
• IPv6

2.6.1 IP Operations
An IP header is placed on the data, which includes information about the destination host and the
ser-vice for which this data is transported. Along with this information, checksum information,
segment number and protocol type is added in the header

When reaching the destination, the reverse operation is performed on the datagram by the
receiving machine in order to retrieve information about checksum (ensures data reliability) and
segmentation (to reorder data).

2.6.2 IP Encapsulation
Figure 2.1 shows how IP encapsulates segments from the upper layers.

---------<TCP Segment or UDP Datagram ------------>

IP Header TCP or UDP Header Application Data

<--------------------------------- IP Datagram------------------------------- >

Figure 2. 1 IP Datagram

Figure 2.1 shows the data for IP header is actually the segment from Transport Layer which itself
includes the TCP/User Datagram Protocol (UDP) header and application data.

8
2.6.3 IP Header
Table 2.1 illustrates the fields contained in an IP header

4-bit IP 4-bit 8-bit 16-bit 16-bit Three 13-bit 8-bit 8-bit 16-Bit 32-bit 32-bit Padding
Version Header type of Total Packet 1-bit Fragment Time- Protocol Checksum Source Destination
Number Length Service Length Identifier Flag offset to- Identifier IP IP Address
Flags Field Field Live Field Address
Field
Table 2. 1 IP Header

The fields are defined as follows:

• IP Version Number: 4-bit, used to identify the IP version (i.e., 4 or 6)

• Header Length: 4-bit, describes the header length and is expressed in multiples of 32
• Type of Service Flags: 8-bit, contain 1-bit flags used to specify delay, priority and
throughput parameters
• Total Length Field: 16-bit, mentions total length of the IP packet in bytes. Valid values
range be-tween 0 and 65535 octets
• Packet Identifier: 16-bit, used to identify the fragment of a datagram
• Flags: Three 1-bit flags, tells whether this packet can be fragmented or not
• Fragment Offset Field: 13-bit, used to measure the fragment offset. It is usually a 64-bit
increment
• Time-to-Live (TTL) Field: 8-bit, senders set this field in range 0-255. It is used to limit the
data-gram’s roaming

• Protocol Identifier Field: 8-bit, used to identify the upper layer protocol, e.g., TCP or UDP
• Checksum: 16-bit, used for error-checking and to inform the destination about size of incoming
traffic. At reaching its destination, the packet is checked against this value and the result is
matched with attached checksum; if the result is the same it means that the datagram has
survived any damage or loss
• Source IP Address: IP address of the sender/source host
• Destination IP Address: IP address of the receiver/destination host
• Padding: IP header should be a multiple of 32-bit, if it is less than 32-bit, extra zeros are attached

2.7 Routing Mechanism

The first step of IP routing is to select the best path for data transport. The question that arises here is
how to select this path. Routing tables are the answer to this question. Routers maintain routing tables
which include information about following

• Directly Connected Networks

• Remotely Connected Networks
• Network Information, i.e., Source of Information, Network Address, Subnet Mask, IP Address of
Next Hop

9
Router reads the destination IP Address from the header of incoming packets and checks it to find out
the destination network address. Then it looks into its routing table and finds an IP address which is the
address of a directly connected device and forwards the incoming packet to that route. This directly
connected device has the information about destination host If the destination IP address is not found in
router’s routing table, then the packet is routed to a default route. Default route is a routing table entry
used to forward packets for which a next hop is not explicitly listed in the routing table.

2.8 Autonomous System

An Autonomous System (AS) is a group of networks and routers controlled by a single administrative
authority AS’s can be categorized in the following types:

2.8.1 Stub AS
The Stub AS is a system which has only a single route, i.e., default route available for data transmission.
This default route is directly connected with another AS. Computers within this AS communicate with
themselves and if traffic is intended to or from outside the stub AS, the default route is used

2.8.2 Transit AS
It is an AS which passes transit traffic between two or several AS. The traffic is not originated nor
terminated in the transit AS. An ISP network is an example of a transit AS. Traffic is not intended for the
ISP; instead it is forwarded to other networks

2.8.3 Multi-Homed AS
It is an AS which has multiple links with one or more ASs. Information is transmitted through only one
link (shortest) and information received through that link is never forwarded back to the same link

2.9 Routing Techniques

Routing can’t route network traffic without the help of routing tables. These tables can be manually or
dynamically maintained Following techniques are used by a router in order to provide routing
functionality:

2.9.1 Static Routing

Administrative authorities of the network statically program routes and update routing tables for each
router in the network. It is a good technique when the network consists of only few connecting devices.
Using this technique, routers can’t intercommunicate to share their routing information. Only the
information provided by administrators can be used for routing and the router forwards traffic only to
pre-defined routes

Static routing is often inapplicable, because organizations tend to have huge networks these days. In
case of link failures, it is the network administrator’s job to specify alternate routes.

10
Therefore, dynamic routing is used instead in order to help routers maintain their own routing tables
and to share routing information with other routers in the network.

2.9.2 Distance-Vector Routing

Routers use algorithms (such as Bellman-Ford) to periodically share routing information to their
immediate neighboring routers in an omnidirectional manner. Each receiver adds its distance value
(distance-vector) to the table and forwards it to its immediate neighbors. The process results in a
cumulative table used by each router to update its own routing table and to get information about the
administrative distance to other network devices

Following protocols use this routing technique

• Routing Information Protocol (RIP)

• Routing Information Protocol Version 2 (RIPv2)
• Interior Gateway Routing Protocol (IGRP)

The process takes a certain amount of time depending on the number of participating devices. There-
fore, in case of a link failure, network traffic may suffer. Often used for big networks, this technique it is
not recommended to be employed in smaller networks because the routing information passed by the
routers may actually have a higher volume than the actual user traffic. The routers are unaware of the
physical topology of the network in distance-vector routing

2.9.3 Link State Routing

In this routing technique, a complex database containing full knowledge about network routers and
information about their interconnectivity is maintained by shortest path first algorithms. Routers share
information about link state (Link Sate Advertisements (LSAs)) to other routers in the network. Instead
of periodic sharing of LSAs, routers wait for events which trigger this process of information sharing
Example of a protocol using this technique is the Open Shortest Path First (OSPF) routing protocol.
Because a huge amount of information is shared by routers, link-state routing can temporarily decrease
network performance significantly. The size of routing tables requires intense memory and CPU
processing

2.9.4 Hybridized Routing

This is a combination of distance value sharing feature from Distance-Vector routing and triggered
invocation of information sharing. By not using periodic information sharing, significant amount of band-
width is reserved for original network traffic

An example of routing protocol designed to use hybridized routing is the Enhanced Interior Gateway
Routing Protocol (EIGRP).

11
2.10 IP Network’s Limitations
IP networks have been serving well until the identification of several key limitations, which prompted
the implementation of MPLS networks. These limitations are:

2.10.1 Scalability
IP networks use two functions, i.e., determination of the route and forwarding of data to that
particular route Forwarding of data directly depends on proper determination of the route. These
days, routers have huge routing tables and these tables increase with every addition of physical
segments to the network.

Consequently, the processing time required for the routing protocols to determine the accurate
route also increases, which makes it harder for the routing protocols to efficiently and quickly find
routes. The larger the size of the routing table, the higher becomes the delay in transmission, i.e.,
poor QoS. Therefore, conventional IP routing doesn’t easily support scalability. There is a need to
either change the routing model or introduce another efficient technique, which, by using existing
IP routing features, provides better QoS.

2.10.2 Class of Service

Conventional IP routing works like a messenger whose only concern is the delivery of the
information but who doesn’t know what kind of message he is going to deliver the delivery service
for all the sources is treated the same. If certain services can be divided in similar groups or classes,
all related services are prioritized on the basis of these classifications. IP routing can enhance the
deli-very process. Services with higher priority can be delivered prior to services having lower
priority.

Unfortunately, the concept of Class of Service (CoS) is not a widely available feature in current IP
routing and therefore cannot be used. Consequently, this situation prompted for the
implementation of MPLS in order to overcome traffic priority issues.

2.10.3 Lack of Route Recovery

Internet has a large physical layout and occurrences of traffic problems are normal. Typical
examples of such problems are route failures. A good infrastructure has built in robustness. If a
route fails, it must be recovered as soon as possible in order to support uninterrupted data
transmissions.

Recovery of an IP route depends on the time it takes to detect the failure, propagating information
to all routers in the network about the occurred failure and the calculation time of new routing
tables Routing protocols are unaware of the exact location of the failed route; they just have
information about the failure and the way they should recover the route.

12
2.10.4 Lack of QoS
IP routing uses a best effort routing mechanism. It uses the best available resources to provide an
“as good as possible” service, but there is no assurance that this effort will meet the expectations of
the clients.

It doesn’t provide QoS even with high bandwidth available and using expensive network equipment.
Therefore, the introduction of a new routing paradigm, as provided by the MPLS technique, be-
comes unavoidable.

2.10.5 Inaccessibility of Bandwidth

Routing may create network congestion and the load balancing mechanism of IP routing is highly
dependent on the size of the network Furthermore, network scalability and load balancing are hard
to implement in classic IP routing. MPLS provides Forward Equivalence Classes (FECs) in order to
overcome the load balancing issue and to access efficiently the available bandwidth which also
results in better QoS for the network services.

13
Chapter-3 Problem Statement

14
Internet has become an essential aspect in everyday life and its resource consumption has in-
creased at a tremendous rate. Engineers around the globe are putting great effort into living up
to consumer’s expectations by developing new applications and algorithms. The increase in
available applications has made Internet into a 24/7 “application” addiction for most of its users
but has also created a negative effect due to high bandwidth requirements which often results
in network congestion.

Internet, since it contains so many networks within itself, has to deal with the increased
sensitivity of its essential components.

Consumers are generally connected through smaller public networks. These networks are often
sensitive to congestion, due to enormous information sharing. Users using applications for
entertainment purposes have to suffer because these applications are usually very sensitive to
packet delays and data loss.

Small networks interact with each other through core networks (i.e., network backbones). The
core networks are highly sensitive to connectivity issues, because unlike smaller networks these
networks have to support large data transfers along large distances; therefore, is the
performance of core networks critical to Internet’s overall performance.

IP networks have generally packet losses and delay. Engineers’ efforts through TE have reduced
these problems to some extent. Engineers’ job is not only to overcome IP network’s drawbacks.
They have to consider issues which highly affect TE and in turn, make the engineer’s job even
more challenging.

3.1 Important Issues

As previously discussed, Internet is a combination of many different networks with respect to
topology, underlying technologies, protocols, applications and standards. Therefore, its traffic
at one particular segment may be different from another.

This project is an effort to draw an image that illustrates existing problems in IP routing and
their solution by using MPLS. The following issues were taken into consideration during project
implementation:

• User’s expectations for reliable, secure and performance-oriented services and systems.
• Emerging different network services with respect to service type, network performance
and targeted user groups.
• Integration of newer and older networks in order to better utilize existing services.

15
3.2 Project Goal
Main goal of the project is to highlight operations of MPLS that enables it to provide better ser-
vices than conventional IP networks. A pre-study of IP routing helped us to understand its main
drawbacks; MPLS features are explained in order to better understand which one technology is
better with respect to network performance, security and robustness. To achieve this goal
smaller objectives were set, as follows:

• Detailed study of IP routing in order to examine its QoS.

• Identification of main problems in conventional IP routing.
• Detailed study of MPLS design, structure and core components.
• Traffic engineering in MPLS and IP networks.
• Identification of main advantages of MPLS networks in comparison to IP networks.

The resulting analysis obtained after the performing the comparison of IP and MPLS networks
will serve as a broad knowledge base in order to prove that MPLS is better in serving
heterogeneous, underlying networks. Students interested in both these technologies may use
this project report for a better understanding of the basics of these network types and identify
their advantages and drawbacks.

16
Chapter-4 Multi-Protocol Label Switching

17
4.1 Background
MPLS is based on the label switching paradigm; on each data packet, a special and independent
label is attached and then with the help of this label packet is switched and routed through the
network. The label is basically a short hand version of a packet header containing information
which helps network nodes processing the label by dispersing and redirecting traffic. X.25,
Asynchronous Transfer Mode (ATM) and Frame Relay (FR) are some examples of existing label
switching technologies within the data communication industry that have been using this
concept for many years.

During the 1990s, many label switching technologies concepts have been merged in order to
improve the performance and QoS of software-based IP routers. Internet Engineering Task
Force (IETF) was charted in early 1997 to harmonize the label switching technologies. Through
these efforts, MPLS emerged as another label switching technology that came into sight with
the main advantage of using the same route and addressing scheme as existing IP protocols.

4.2 MPLS Concepts

MPLS is an advanced technology that uses service incorporation between Layer 2 (Data Link
Layer) and Layer 3 (Network Layer) which enables traffic flow control in the network; it is called
Multi-Protocol because it can be used by multiple layer 3 protocols. MPLS is built by IETF to
enable a new routing architecture and to facilitate fast and scalable Internet communication. To
gain reliable distribution between different links, MPLS enables the network operator to
reallocate packet flows through TE.

Basically, MPLS is a contemporary technique which accelerates the transfer of data networks. In
MPLS packets are utilized for accelerating data transfers; first these packets are encapsulated
by assigning labels at MPLS network ingress points and then the packets are quickly transferred
through the network on so-called label switched paths

4.3 MPLS Applications

Most important applications of MPLS are:

4.3.1 Traffic Engineering

All packet traffic between two points is sent over the shortest path using traditional IP routing
even though multiple paths may exist. This concept is illustrated in Figure 4.1. In case of high
traffic volumes, traffic congestion may easily occur on some routes.

18
MPLS employs the residual bandwidth more efficiently by allowing packets to be routed along
unambiguous routes and with definite bandwidth guarantees; this is in contrast with rather
adding more bandwidth to handle the high traffic as is the case with conventional IP routing.

Figure 4. 1 Traffic Volumes in Conventional IP Network

Label Switched Paths (LSPs) connections can be optimized and preempted through MPLS TE.
Characteristically MPLS TE is organized at the core of the MPLS network, whereas QoS is used at
its edges; this approach ensure that high precedence packets are routed first while TE suitably
manages the available bandwidth resources and avoids network congestion. The need of TE
originates from the necessity of identifying required routes in a dynamic manner, due to the
random nature of Internet

4.3.2 Path Management

Path Management carry out the entire task related to maintenance and instantiation of LSP
tunnels. LSP is used in association with the MPLS signaling protocols. Path Management
characterizes the rules for path selection criteria and supporting the established LSP tunnel.
Path placement and path selection are the basic elements of Path Management. Path
placement define LSPs through signaling protocols which also works as label distribution
protocol whereas LSP tunnels are enabled by path selection at the tunnel origin

4.3.3 Virtual Private Network

Virtual Private Network (VPN) encapsulates traffic between two or more devices over Public
Shared Network (PSN), when these devices are not on same private network. The main
advantage of VPN is building safe and secure connections over public networks through remote

19
locations without any operational cost incurred by dedicated lines. MPLS permits VPNs through
a connection-oriented framework, allowing carriers to organize VPNs over IP infrastructure

VPN is an elegant substitute of Frame Relay Permanent Virtual Circuits (FRPVCs) which is
sustained by MPLS. The main advantage of MPLS VPNs model over FRPVCs is that VPNs are
highly scalable. From an end user point of view, another advantage of MPLS VPNs is that
simplified routing is possible when compared to the FRPVC model since MPLS VPNs can use the
default route as service provider while managing routing PVCs oblige topologically composite
back-bones. VPNs also convey the QoS to their consumer through emerging Differentiated
Services (DiffServ) techniques.
According to such QoS requirements, the traffic that flows with DiffServ through the network
backbone is split into different traffic classes. The traffic belonging to the-se classes is
differentiated through specific header bits and different labels. On the basis of the bits set in
the header and the label types, routers take forwarding and queuing decisions in or-der to
sustain the particular QoS.

4.3.4 Multi-Protocol Support

MPLS have attractive features to support multiple protocols. Internet is basically a mixture of
dissimilar technologies such as ATM, IP and FR; in these circumstances MPLS cannot only be
used together with all these technologies but can be also used with other network
technologies. Because the general nature of multi-protocol support not only expands the QoS
of the existing network technology but can also provide the solution to make the best use of
the existing resources. The possibility of mixing different network technologies is in this context
fascinating

4.4 MPLS Header

The MPLS header is 32 bits in length. It is inserted between the layer 2 (DLL) and the layer 3
(network) header. It is assigned at the MPLS ingress router. Every incoming packet is assigned a
shim header (if no existing labeling technology could be used). The layout of the MPLS shim
header is depicted in Figure 4.2

Figure 4. 2 MPLS Shim Header

20
4.5 MPLS Operations
MPLS operations are executed on Label Switch Routers (LSRs) and Label Edge Routers (LERs). A
LER, also called ingress router, works as the main interface between the MPLS network and
existing layer 2 technology. This router works at the boundary of an MPLS network whereas its
interfaces are associated to other networking technologies. LER sends packets by attaching la-
bels at the MPLS ingress, when it receives packets from outer networks

Figure 4. 3 MPLS Operations

4.6 MPLS Network Architecture

The MPLS architecture is a combination of MPLS routers linked through mesh topologies. The
MPLS architecture consists of two components and two planes:

• Forwarding Component

21
 • Control Component
• Control Plane
• Data Plane

4.6.1 Forwarding Component

This component only forwards data packet by fetching the appropriate label information from a
database maintained by the Label Switched Router (LSR).

4.6.2 Control Component

The control components create and maintain label information that enables information
transfer between consistent groups of labels. In MPLS, data transmission takes places on Label
Switched Paths (LSPs). By using the Resource Reservation Protocol (RSVP) and the Border
Gateway Protocol (BGP) the labels are exchanged among routers.

4.6.3 Control Plane

The control plane exchanges information between adjacent devices through different protocols
such as Open Shortest Path First (OSPF), External Interior Gateway Protocol (EIGRP), Interior
Gateway Protocol (IGRP) and BGP. Label information is also exchanged in the control plane
through dedicated label exchange protocols (i.e., RSVP, Label Distribution Protocol (LDP) and
Tag Distribution Protocol TDP)

4.6.4 Data Plane

Data plane forwards the packets on the basis of their label. The packets are redirected through
label look ups, located in the Forwarding Information Base (FIB) Table. FIB table is generated
and maintained through LDP and TDP

4.6.5 Label Switching Router

Label Switched Router (LSR) is a device which resides somewhere in the middle of the MPLS
network. Its responsibility is to switch the labels to transmit packets. It is often known as a
transmit router.

When LSR receives a packet, it reads packet header and uses the included label to determine
the next hop. It removes the old label from the packet header, attaches a new label and routes
the packet

4.6.6 Label Edge Router

22
The Label Edge Router (LER) is positioned at the edge of an MPLS network. At the ingress, it
receives IP datagrams, determines appropriate labels to be affixed, labels the packets and then
forwards the labeled packets to next hop in the MPLS network. If a packet is destined to exit
the MPLS network, LER strips off the labels and forwards packets using conventional IP routing
techniques

4.6.7 Label Switching Path

A Label Switch Path (LSP) is a path from the Ingress LER toward the Egress LER. In short, an LSP
creates a connection-oriented technology because the path is associated based on the
precedence of the data flow. LSP describes the route for a Forwarding Equivalence Class (FEC)
that comprises a set of LSRs that a packet must traverse consecutively in order to reach the
destination. On the basis of the core technology, a connection is setup instead of the
requirements of the data flow

Figure 4.4 illustrates the flow from ingress LSR, the LSP tracked by three transitional LSRs and
egress LSR.

23
 Figure 4. 4 LSP through a MPLs Network

4.6.8 Label Distribution Protocol

Label Distribution Protocol (LDP) is responsible for label management. This protocol is discussed
in more detail in Chapter 6: Label Management.

4.6.9 MPLS Label

In MPLS, labels are attached to IPv4 and IPv6 packets. These labels are attached between the
network layer header and data link layer header. A label consists of 4 or 8 bytes.

Figure 4.5 shows the fields of an MPLS header

0 1 2 3 4 5 6 7
Label

CoS S
TTL

Figure 4. 5 MPLS Label Format

• Label: a 20-bit field which actually contains the label value. It also provides
information needed to forward the packet.
• CoS (Class of Service): a 3-bit field used in setting the schedule and to discard the
algorithms which are applied to a packet during its transmission.
• S: a 1-bit field representing Bottom of the Stack, If it is set to 1, it means it’s the last
entry in the label stack otherwise it is set to 0 for all other label stack entries.
• TTL: an 8-bit field used to encode a time to live value.

4.6.10 Forwarding Equivalence Class

The packets that have the same traffic characteristics and priority requirements are
encapsulated to the same Forwarding Equivalence Class (FEC). LER assigns the FEC once,

24
when the packet enters the MPLS network. The packets that have the same FEC are
redirected on the same path, in the same manner. Before redirection, the specific packet FEC
is encoded within the label and inserted into the packet.

As long as the packet remains in the same MPLS domain, there will be no extra processing of
that packet. But, in order to decide the next hop, the LSR uses the label and recognizes the
FEC index in its table. When the packet is redirected to the next hop the old label is removed
and a new label is dispatched on that packet, because in some cases packets belonging to
same FEC may not be bound to the same label and may have different labels

4.7 MPLS Operation Modes

4.7.1 Frame Mode

In MPLS frame mode, routers exchange the pure IP packets with labeled IP packets. Figure
4.6 demonstrates how MPLS frame mode label allocation and distribution take place.

25
 Figure 4. 6 Frame Mode MPLS Label Assignment and Distribution

According to the information available in Figure 4.6, edge LSRs R1 and R4 are connected with
LSR R2 and R3. LSR allocates a local label (172.16.10.0/24) and then broadcasts this label on the
upstream path, after LDP neighbor formation and IGRP convergence. Consequently, the FIB, LIB
and LFIB (the control and data structures) are populated with proper values.

Edge LSR R1 broadcasts the same upstream path information to LSR R2 after assigning the local
label (implicit null). Similarly, LSR R2 and LSR R3 broadcast the upstream path to the network
destination after assigning the local labels L2 and L3, respectively. The label distribution can
either be spontaneous downstream or on demand downstream.

The illustration of data packets sent to the destination network (172.16.10.0/24) through the
MPLS domain is depicted in Figure 4.7 shows that the edge LSR R4 broadcasts the labeled pack-
et to the downstream LSR R3 after imposing the label L3. Here LSR R3 exchanges the ingress
label L3 for the egress label L2. On LSR R2 ingress label L2 plots on the implicit null label.

Therefore, LSR R2 forwards the IP packet to the Edge LSR R1 as shown in Figure 4.7

26
 Figure 4. 7 Frame Mode MPLS Forwarding

By using protocol/type field in the frame header the routers can determine the payload type of
a received frame. For example, in case of Ethernet IEEE 802.3, the 13th and 14th octets of the
packet consist of IEEE 802.3 Length field or IEEE 802.3 Type. The values in these octets help
routers to determine the frame as either a conventional IP packet or a labeled IP packet

4.7.2 Cell Mode

In Cell Mode, the MPLS domain uses the label information from the existing layer 2 technology.
ATM’s VPI and VCI values are used to forward the labeled ATM cell in the MPLS domain. The
cell mode is illustrated in Figure 4.8.

Figure 4. 8 Cell Mode MPLS Label Imposition

According to Figure 4.8, the cell mode MPLS network consists of an egress edge ATM LSR, in-
gress edge ATM LSR and core ATM LSRs. In the MPLS domain, the LSR interface carry pure ATM
cells, called Label Switching Controlled ATM interfaces (LC-ATM). In the implementation of in-
tegrated ATM LSR, LC-ATM carries both control plane and data plane packets.

27
 .

Figure 4. 9 Cell Mode MPLS Domain

For the Interior Gateway Protocol (IGP) and LDP packet exchange, control plane adjacency is
essential among neighboring switches as illustrated in Figure in 4.9. IP adjacency is formed by
an ATM Edge LSRs and ATM LSRs which ensures (through an ATM switch) suitable forwarding of
the IP routing information. On all ATM switches the MPLS control VC is available by default.

Cell Mode MPLS Operation

The top label in the label stack is encoded as a virtual circuit of VPIs and VCIs in order to permit
MPLS forwarding in the ATM domain. The top label in the label stack is interleaved among the

IP and the ATM header. Data plane packets are redirected by using this mechanism whereas
control plane packets are replaced on VC.

28
 Figure 4. 10 Cell Mode MPLS Label Allocation and Distribution

Figure 4.10 illustrates this concept. Edge LSR R1 is directly connected to the 172.16.10.0/24
network where R1 and R2 are Edge ATM LSRs, and where A1 and A2 are ATM LSRs. In Cell
Mode MPLS operation, the process of label allocation and distribution is similar to the one in
Frame Mode MPLS operation. Downstream Router R1 broadcasts the same upstream path by
allocating a label to destination network 172.16.10.0/24. On ATM LSR A1 and A2, this process is
repeated.

The only difference between Cell Mode MPLS and Frame Mode MPLS is that Cell Mode MPLS
uses interface level label space whereas the Frame Mode MPLS uses both interface level and
router level label space

4.8 Loop Detection and Prevention

The MPLS network routing protocols are mainly based on distributed computation, so until
convergence occurs, the routing protocols normally may have transient routing loops. Similarly,
the MPLS switched paths may also form loops unless preventive measures are taken. Routing
loop prevention is reduced with the help of Time-To-Live (TTL) field in the IP header, in the case
of IP routing. Packet is discarded when the TTL value reached zero, because at every forwarding
hop, the value of TTL is decreased by one.

MPLS packets forwarded on ATM labels have no such type of loop prevention methods since
ATM cell does not contain the TTL field. As long as the switched path exists, the ATM traffic re-
mains in a routing loop which will severely impact the performance of the network.

29
Prevention and Detection are two methods used for handling the routing loops in an MPLS
network. In Prevention, different methods are used to stop the routing loop from ever being
formed on the LSP. In the Detection method, the routing loops may initiate but the MPLS map-
ping distribution message will ignore or stop forwarding to that LSP by examining the loop [7,
12].

4.8.1 Frame Mode

Labels are connected to specific FECs in an autonomous control mode when MPLS is running in
Frame mode. Label assignment to precise FECs is done in MPLS only when FECs stay alive in the
routing tables of LSRs. In this way, MPLS can reveal and prevent forwarding loops when the LSP
is established.

4.8.2 Frame Mode: Control Plane Loop Prevention

In a conservative IP network, Interior Routing protocols prevent routing loops. Routing loop
detection is a very important function from a network performance point of view and this
function is done by LSRs in the MPLS network. Prevention of forwarding loops must occur in the
control plane since the LSP is created within the MPLS control plane.

In frame mode, the LSRs use the same Interior Routing protocols for populating and building up
the routing tables as those used in conventional IP networks. Frame mode lets the LSRs to share
the routing table information and consequently the prevention of routing loops becomes
possible

4.8.3 Data Plane Mode

In MPLS networks the Data Plane mode is used to prevent the routing loops by monitoring the
TTL field of the incoming IP packet. The value of TTL field is decreased by 1 in the MPLS domain
but if the value is already 0 then, in order to prevent forwarding loops, the IP packet is dis-
carded

30
 Figure 4. 11 Loop Detection Processes

4.9 LSR Mode

There are the following LSR Mode types:

• Label Distribution Mode

• Label Retention Mode
• Control Mode

4.9.1 Label Distribution Mode

LSR is responsible for the distribution of the labels. Label distribution is further discussed in
Chapter 6 Label Management.

4.9.2 Label Retention Mode

Label Retention is the responsibility of the LSR. This is also further discussed in Chapter 6 Label
Management.

31
4.9.3 Control Mode
Structured and self-governing FEC are performed in Control Mode. LSR creates the FEC local
binding according to FEC without involving other LSRs. FEC is bound to a structured LSP unless
egress LSR identifies the FEC or label binding from an approaching hop

4.10 MPLS Packet Forwarding

There are different methods of packet forwarding in MPLS compared to conventional IP packet
forwarding. These methods are:

1. Push, Pop and Swap are three major operations performed in MPLS label forwarding.
When label packet is received at LSR either top label should be swap, pop or push and is
determines by LSR according to LFIP information. These three operations are as follows:
• Swap: The label is swapped by a new label and the path associated with new la-
bel is used for packet forwarding.
• Push: A new label is pushed on top of the existing label. It helps encapsulating
the packet in another layer of MPLS. This operation is used in MPLS VPNs.
• Pop: The label is stripped off the packet. It may result in decapsulation of an in-
ner label. At this point, the packet may be leaving the MPLS network.

32
Chapter-5 MPLS Traffic Engineering

33
5.1 Overview
Internet community is facing big challenges with the rapidly growing usage of real time
applications. Most service providers are facing several problems in their networks such as data
congestion, packet delay and loss. Consequently, they need special tools and techniques in
order to make their networks more efficient with such an unpredictable traffic flow.

Traffic Engineering (TE) is a technique which optimizes the performance of a

telecommunications network. TE is used for analyzing, predicting and modifying the behavior of
data transmission over that network. TE is also known as traffic management.

TE has the ability avoid traffic flow congestion and to dynamically define network routes. When
data traffic gets congested at nodes, MPLS TE manages the data traffic between the congested
nodes. MPLS helps TE to find the shortest path and also helps in utilizing unused paths. MPLS TE
is not only aware of the flow of each data packet but also takes the responsibility of properly
redirecting the traffic flow and fulfills the traffic QoS requirements.

5.1.1 Basics of TE
During data transmissions there is a possibility of losing data packets. TE has the basic function
of avoiding this loss by efficiently using the available bandwidth through effective techniques
and to maintain traffic flows.

In traditional IP networking the main role of TE is to save bandwidth in an effective way.

Secondly, TE tries to avoid congestion of data traffic in the network nodes and to maintain the
traffic flow between a source and a destination.

Figure 5. 1 IP Forwarding Networks

34
Figure 5.1 illustrates this concept. In this simple example, the network topology contains links
with the same cost. Assume R5 as destination and R1 as source. In this case, there are two poss-
ible ways of transferring data from the source to the destination, i.e., R1 to R5. The first path is
R1 to R2 and then R5. The second path between R1 and R5 is from R1 to R3 followed by R4 and
finally R5. As we mentioned, all the links have the same cost; from a cost point of view, the
second path: R1 -> R3 -> R4 -> R5 is more expensive. Consequently, in order to use the available
bandwidth more effective, a static route may be employed for R1 -> R2 -> R5

5.2 MPLS-TE
In MPLS TE, source-based traffic routing is introduced since conventional IP routing may incur
routing inefficiencies due to its mechanism of destination-based routing. If we examine most
networks using destination-based routing, we will observe that we will be unable to utilize the
full benefits of this type of causal deployed resources. In MPLS, TE overcomes these types of
deficiencies by efficiently utilizing the available bandwidth and link capacity. Still we need to
solve the congestion issues but in the case of MPLS TE we have access to the best solutions to-
wards congestion avoidance problems at the network nodes.

5.2.1 Overview
To efficiently use the available bandwidth in the MPLS networks, TE implies configuring,
computing and calculating the networks paths. MPLS TE is capable of using the bandwidth more
efficiently of the existing networks path. MPLS TE can be used as a tool for achieving and
implementing the most cost valued, automated and differentiated services in a network.

MPLS TE mainly addresses performance issues and optimizations of operational IP networks.

Since TE has a much broader range of QoS logic, it tends to maximize the operational efficiency
of the network. It has the ability to redirect traffic to those network nodes where the links are
under-utilized while assessing, if necessary, the QoS problems, traffic distribution. In this way,
TE helps minimizing the network congestion. TE can be automatically available in circuit switch-
ed networks by using offline tools.

5.2.2 Load Balancing

An example of TE network architecture is depicted in Figure 5.2. The architecture highlights the
basic requirements having three transatlantic links where each link has a different capacity. As
known, a conventional IP routing scheme will not distribute the traffic load across each path
and data traffic cannot be transferred equally across unequal paths.

35
 Figure 5. 2 Shortest Path Computations

In Figure 5.2 data between A and E can be equally transferred crosswise, from top and bottom
paths, but this does not catch the different link sizes between [B-D] and [C-D] as shown in the
figure. In the top path, the available bandwidth between A and B is 10Mbps and between B and
D is 60Mbps, which is different to bottom path where bandwidth between A and C is 10Mbps
and between C and D is 40Mbps. The need in this case is to utilize both these paths efficiently
so that there is no wastage of network resources

5.2.3 Link Congestion

Main problem in conventional IP routing is that in order to calculate the best IGP paths, many
such paths may be over deployed. Furthermore, alternative paths remain often underutilized or
sometimes even not utilized at all.

36
 Figure 5. 3 Over-Utilized Links

Figure 5.3 illustrates this mechanism. The link F-E-D-B is not used for forwarding traffic from G
to A. From this figure, we notice that the traffic follows the shortest path to reach the
destination via F-C-B due to shorter number of hops. Meantime, by adding more bandwidth,
IGP can be determined to take a path with more hops for the Shortest Path First (SPF)
calculation. Nevertheless, this computation is considered to be difficult since it has no concept
of traffic congestion

5.2.4 Link Protection

In a primary LSP, if there is an issue such as a link going down, it means that there is a path or
device failure. In order to forward the traffic again, the routing protocols have to perform the
full SPF computation again. This can take several seconds because such computations are a
time-consuming process.

37
 Figure 5. 4 Primary Path Failures

Such scenario is illustrated in Figure 5.4. If there are any traffic issues in the direct link from A to
D, before traffic can cross C, the routing protocol must recompute the route. In this scenario,
the problems described in Figure 5.2 are easily resolved through MPLS TE. TE builds a tunnel
(the LSP) all the way through the MPLS core.

Any packet entering at the beginning of the tunnel follows the same path throughout the
tunnel until it exits at the other end. Two protocols are used in such cases: Constraint Based
Routing (CBR) and RSVP. Both these protocols are used for label distribution within the MPLS
domain.

5.3 Advantages of MPLS-TE

MPLS TE provides efficient, consistent, technical measurements, categorization, modeling and
management of network traffic. MPLS provides basis for TE to ISPs for the following reasons.

1. Continuous support for definite paths and allowing the network to identify correctly the
exact path that an LSP takes across the MPLS network.

2. Individual LSPs used to identify traffic jams and trunk utilization.

3. CBR gives the ability to join LSPs together and meeting thus specific traffic requirements.

4. Solutions provided by MPLS to ATM networks can be easily adapted to run on packet
switched networks.

5. ISPs can balance the traffic in their networks allowing thus for a better utilization of their
hardware resources such as routers and switches. Thus, all the network components can
provide better services.

6. TE makes sure that no network component is over- or underutilized.

7. TE allows the available bandwidth to be used very efficiently. Even in case of network con-
gestion, the network resources are allowed to get the best performance.

5.4 MPLS-TE Operations

In MPLS-TE there are several types of operations.

38
5.4.1 Distribution of Link Information
It extends the IP link state topology information distributed by the LSR compared with
restrictions-based routing. It allows a growing list of links with known attributes and constraints
in order to perform the path selection algorithm. OSPF and IS-IS are examples of two
fundamental routing protocols that can be employed for the distribution of link attributes for
the LSR TE database development. MPLS TE bandwidth attributes are divided into eight priority
interval levels sets for the LSP TE. TE metric attributes are used to link the same paths in the IGP
metric in order to optimize the administrative group maintenance, including various special
rules.

5.4.2 Computation of Path

The TE topological database is developed by TE LSP alongside with the SPF algorithm. They
integrate with each other in order to implement the Constrained Shortest Path First (CSPF)
algorithm. This is used in order to find the shortest and the optimal path. For network resources
there are no guarantees of optimal mappings of the traffic stream.

5.4.3 LSP Signaling

MPLS TE signals LSP through RSVP through several types of messages:

1. LABEL REQUEST: used to attach the label at each hop.

2. LABEL: used for RSVP message distribution.

3. EXPLICIT ROUTE: defines the hop list for signaling

4. RECORD ROUTE: collects label and hop information during signaling

5. SESSION ATTRIBUTE: defines the requirements of LSP attributes such as protection, priority.

5.5 MPLS-TE Classes

There are two classes used in MPLS-TE in order to improve network performance.

5.5.1 Traffic Oriented

In this class, emphasis is set on the improvement of QoS in the traffic streams. In a best effort
service model, the following objectives are considered for improving network traffic
performance in a single class.

39
 • Minimizing packet loss ratio
• Minimizing the delay
• Maximizing the throughput

5.5.2 Resource Oriented

In this class, the main objective is to optimize network resource utilization. It’s often observed
that in traditional IP routing, one particular section of a network may easily become over
utilized while others remain underutilized. Moreover, there may be cases when the network
cannot offer enough resources in order to serve end users and applications, e.g., when network
traffic is required to follow some predefined paths that result into over utilization of certain
network segments. TE resource-oriented class is intended for the best usage of such resources
so as to overcome such scenarios and optimize network resource utilization.

5.6 RSVP and TE

RSVP is used in MPLS TE for prior reservation of the network resources before data
communication takes place. RSVP is a signaling protocol and was developed by IETF’s Integrated
Services (IntServ) working group. It provides end-to-end guarantees by provisioning the
required band-width for data transfers allowing thus for minimum delays to be achieved.

RSVP also provides information about the QoS requirements. It can be used in concert with
other IP routing protocols in order to reserve resources which results in enhanced network
performance.

5.6.1 RSVP Packet Format

RSVP packet format is defined in Table 5.1. The corresponding header fields are defined as
fol-lows:

4 4 8 16 16 8 8 32 15 1 16
Version Flags Type Checksum Length ReservedSent Message Reserved MF Fragment
TTL ID Offset
Table 5. 1 RSVP Packet Format

• Version: a 4-bit field which indicates protocol’s version number (current version is 1).
• Flags: a 4-bit field (currently no flags have been defined).
• Type: an 8-bit field which contains one of the 6 following and possible (integer)
values shown in Table 5.2
• Checksum: a 16-bit field which represents the standard TCP/UDP checksum covering
the contents of the RSVP message.

40
 • Length: a 16-bit field which represents the length of the RSVP packet in bytes,
including header and the variable-length payload that follows. If the More Fragment
(MF) flag is set or the fragment offset field is non-zero, this is the length of the
current fragment be-longing to a larger message.
• Send TTL: an 8-bit field which indicates the IP TTL value for the message.
• Message ID: a 32-bit field used when fragmenting a larger message.
• More Fragments (MF) Flag: Low-order bit of a 1-byte word with the other 7 high-
order bits specified as reserved. MF is set for all but the last fragment of a message.
• Fragment Offset: a 24 bit field which represents the message fragments as a byte
offset.

Table 5.2 comprises the RSVP message types.

Value Type of Message

1 Path
2 Reservation-Request
3 Path-Error
4 Reservation-Request Error
5 Path-Teardown
6 Reservation-Teardown
7 Reservation-Request acknowledgement

Table 5. 2 RSVP Message Types

5.6.2 RSVP Object Fields

Table 5.3 contains information about the RSVP object fields. These are defined as follows:

16 8 8 Variable in Length
Length Class-Num C-Type Object Contents

Table 5. 3 RSVP Object Fields

• Length: a 16-bit field which contains the total object length in bytes (it should always be
a multiple of 4 with a minimum value of at least 4).
• Class-Num: used to identify the object class, where each class has a name. An RSVP
implementation must be able to recognize these classes.
• C-Type: class object type, unique within the Class-Num. Its maximum length is 65528
bytes. Class-Num and C-Type fields (together with the flag bit) are used as a 16-bit
number which uniquely defines a type for each object.
• Object Contents: The Length, Class-Num, and C-Type fields specify the form of the
object content.

41
5.7 RSVP Message Types
In order to remove the reservation information, to give confirmation reports, error messages or
to maintain various data flows, RSVP uses different types of messages, as follows.

5.7.1 RSVP Path Message

It is a periodic message transmitted by the sender in order to refresh the state of the reserved
path. Routing protocols are used to transmit this message along downstream. The path used in
the flow of the path messages is the same as the one for the reserved flow of application data.
It helps maintain the path state and lets a router identify adjacent nodes required for that
particular session. Consistent path state is important in the transmission of reservation request
messages.

5.7.2 RSVP Path Tear Message

This message is transmitted by the sender in order to remove the path state from the routers.
The path is maintained with the help of path messages. In order to improve the utilization of
the available network resources, it is of crucial importance to release any reservations no
longer in use. Path tear messages serve this vary purpose and result in quick recovery of
network resources.

5.7.3 RSVP Path Error Message

It is transmitted by the receiver in order to inform the sender if there are any parameter
problems in the received path message. It is not a frequent or a periodic message. This message
is only sent in case of path reservation problems.

42
 Figure 5. 5 Path Message and Path Error Message

Figure 5.5 illustrates the mechanism of how the path message is transmitted and the feedback
received in case of path errors.

5.7.4 RSVP Reservation Request Message

Reservation-Request message follows the same path as the RSVP path message. It is sent by the
receiver after receipt of a path message. It is used to reserve the necessary resources for data
transmission. RSVP carries this message.

5.7.5 RSVP Reservation Confirm Message

It is sent by the sender upon receipt of a reservation request message in order to confirm that
all the required network resources have been reserved. At the receipt of this message, the
receiver is allowed to know that all the required network resources are reserved for the
required data transmission and now is the time to transmit.

5.7.6 RSVP Reservation Tear Message

Reservation tear message is sent by the receiver in order to remove the resource reservations
that were reserved as a result of reservation request message. It is important to release all
reserved network resources which are not in use anymore.

5.7.7 RSVP Reservation Error Message

In case of problems occurring during resource reservation of the required network resources, it
is important to inform all the communication parties about the problem. It is sent by the
intended receiver of application data.

43
 Figure 5. 6 Reservation Request Message and Reservation Error Message

Figure 5.6 illustrates the reservation request message sent from the receiver in case of resource
reservation problems. The flow also describes how the reservation error message is transmitted
by the receiver of application data.

5.8 RSVP Reservation

Resource reservation is a very important mechanism in MPLS. It depends on the information on
the state of the network path and is used for reserving an appropriate path for data
communication. The following key functions are performed during RSVP reservations.

5.8.1 Sender Selection

Sender can be selected in two ways depending on the situation.

• Explicit Sender: a list of the nodes intending to start data transmission is explicitly
created.
• Wildcard Sender: all network nodes are marked as senders. The device wanting to act as
sender can later on confirms itself as sender.

5.8.2 Session Reservation

Resources can be reserved either distinctly or shared. During shared reservations, sessions are
used to allocate shared resources amongst participating nodes.

5.8.3 Reservation Styles

There are three reservation styles used during resource reservation, as follows:

• Fixed Filter (FF): In this reservation type, distinct reservations are made between a
sender and a receiver. These reservations are not shared between others possible
participants. Video applications and applications using unicast communication are
typical examples of this style of reservation.

44
 • Wildcard Filter (WF): In this reservation type, all wildcard senders share a single
reservation regardless of the number of senders sharing it. However, at any given time
instance only a few senders are allowed to transmit.
• Shared Explicit (SE): In this reservation type, a receiver allows explicit senders to share a
single resource reservation. A given bandwidth share is set for a group of senders to be
used during transmission.

5.9 Quality of Service

The ability of providing superior (value-added) services to network traffic through the use of
various technologies such as MPLS, ATM and SONET is often referred to as Quality of Service
(QoS). Its keys governing principles are to reduce packet loss, jitter, bandwidth latency, and to
correctly prioritize network traffic. These prioritizations do not mean quality degradation of
another network traffic.

The QoS is said to be satisfactory to a service if it meets the terms stated in the Service Level
Agreement (SLA) with a direct result of customer satisfaction in the received service. The aim of
QoS is to ensure highest possible throughput, reduce the packet loss, jitter and delay and allow
for enough capacity for the network traffic.

The link “cost” is also an issue in QoS, since it is needed to ensure minimum usage of the avail-
able resources without reducing the quality of the offered services. In traditional IP routing, two
fields in the IP header allow for limited provisioning of QoS in IP: Type of Service (ToS) and Class
of Service (CoS). These fields allow for a rough classification of the network traffic flow.

5.10 Internet Service Models

Existing Internet service models are often categorized in four categories, as follows:

• Best-Effort Service Model (no QoS guarantee)

• Models with QoS Guarantee
 Integrated Services Model
 Differentiated Services Model
 Hybrid Service Model

5.10.1 Best-Effort Service Model

Best effort service model does not allow for any “promises” on whether the traffic will be
delivered or when. In this case the sender has no information about data delivery. It tries to

45
offer the best service it can and is using the cheapest paths. As there is no resource allocation in
this model, this type of traffic has no guarantees on data delivery status, i.e., no information if
the packet is corrupted, delivered or lost.

Routers use a first-in-first-out (FIFO) mechanism when forwarding the best-effort packets.
There are no promises of QoS while the main focus in this model is on the delivery and not on
the time instance. Real time applications such as video and voice streaming use this service
model, because data delivery is more important for these applications than the time of deli-
very. By analyzing the performance of the best-effort model it can be concluded that this ser-
vice model is not suitable for applications requiring precise QoS levels.

5.10.2 Integrated Service Model

The Integrated Services (IntServ) model is used in combination with RSVP and allows for a flow-
based service state at every hop. Whereas IntServ provides QoS guarantees, this model is very
complicated and if used for providing telephony services to subscribers it may result in wrong
billing information.

There are also known scalability problems in this model since the routers need to uphold state
information at each hop. Large heterogeneous networks such as Internet cannot easily provide
the required QoS for this model.

Components

There are three main components of IntServ model

• Class of Service (CoS)

• Control Mechanism
• Resource Reservation Protocol

Class of Service (CoS)

There are three classes of service

1. Best Effort: This class is used for text-based application. It is, as the name implies, just
best effort which also means that there is no QoS guarantees on the traffic flow.
2. Guaranteed: This class is used for real time applications. It allows for precise levels of
bandwidth, delay and maximum jitter required for a certain traffic flow.
3. Controlled Load: This CoS provides a constant traffic flow level but with no guarantees
for QoS.

46
Control Mechanism

There are traffic classes on which the traffic is divided into flows and in each class all types of
traffic can be combined. For each class, suitable control mechanisms are used. Several
important control mechanisms are used in IntServ:

1. Token Bucket Filter (TBF): In order to guarantee the necessary QoS, TBF uses queue
allocation and the available bandwidth such that the highest jitter limit, delay
constraints and required bandwidth can be met.
2. Resource Reservation Protocol (RSVP): It is used to reserve buffer space and bandwidth
for each traffic flow.
3. Random Early Detection (RED): When the buffers of routers become full RED assists in
reducing in a fair way the queued packets.
4. Weighted Fair Queuing (WFQ): Another mechanism employed for buffer management in
routers.

5.10.3 Differentiated Service Model

Differentiated Services (DiffServ) model is used to supervise the resources of each network
segment (because IP is an overlay network, consisting of many segments). In order allow for a
QoS behavior, each packet is “colored”. For aggregated flows, it provides QoS guarantees. This
model is often used with TE and MPLS. However, it does not provide an end to end QoS
guarantee. In this model, the traffic is treated only as a base class. For applications requiring
end to end QoS, this model cannot be used.

It is quite obvious that voice applications should have dedicated circuits in order to ensure the
required QoS since in this case there would be no noticeable delays. In this case the question
arises why we use QoS? A simple answer is to provide good services, but it is a complicated task
which has many issues to be taken into consideration.

5.10.4 Hybrid Model

It is considered to be a very powerful emerging model but unfortunately it is still under study
[24]. It is supposed to take into consideration the advantages of both the IntServ and DiffServ
models. It will most likely eliminate the deficiencies existing in these service models while, in
the same time, will merge their most powerful features.

47
Chapter-6 Label Management

48
MPLS is highly dependent on efficient label binding. In order for MPLS to work efficiently, the
label distribution mechanism must be both efficient and reliable. LSRs are responsible for
routing the traffic with the help of these labels. Labels are usually exchanged between LSRs
through the use of already existing routing protocols (piggyback). For instance, IGRP is an
example of such a routing protocol.

MPLS supports multiple methods for label creation and binding, i.e., how to map layer 3 to
layer 2 information, and thus establishing LSPs. MPLS employs mainly three methods:

• Topology Based Method

• Request Based Method
• Traffic Based Method

6.1 Label Distribution Protocol

The Label Distribution Protocol (LDP) is an important protocol within the MPLS architecture,
which performs label distribution. It helps LSRs to exchange label mapping information and
enables them to establish proper LSPs in order to forward network traffic.

LDP performs distribution of labels between directly connected LSRs (peers). For label
assignment, it is very important that one LSR inform its adjacent LSR about amendments
performed to the label. This must be done in order for all LSRs to know what needs to be done
with the labels such as to ensure an accurate selection of LSPs.

LDP delivers not only stream mapping information but also distributes the labels across the
MPLS network. LDP uses piggybacking as label distribution mechanism. LDP does not perform
well when using ERP but it works under BGP. This is mainly due to the fact that BGP is a multi-
protocol and more competent than ERP. BGP carries labels with ease. An LSR is key device in

49
LDP since every time new traffic comes to an LSR; the LSR first forwards the label information
and then forwards corresponding traffic.

6.1.1 LDP Packet Format

Version PDU Length

LDP Identifier
LDP Message

Table 6. 1 LDP Packet Format

Table 6.1 illustrates the LDP packet format. In LDP packet format

• Version: 2 bytes long, indicates the current version number of the LDP protocol
• PDU Length: 2 bytes long, indicates the total packet length of LDP message

• LDP Identifier: 6 bytes in length, used for packet detection by LSR. The first four bytes in
the LDP identifier field are used to encode the IP address of the LSR whereas the last
two bytes are used for LSR labeling.
• LDP Message: defines the type and length of the LDP message.

6.1.2 LDP Message Format

The LDP message format is given in Table 6.2.

U Message Type Message Length

Message ID
Parameters

Table 6. 2 LDP Message Format

• U: denotes unknown message. Receiver will discard the message if the bit is set.
• Message Type: denotes the type of message.
• Message Length: denotes length of message ID in octets. Mandatory and optional LDP
parameters are controlled by this field.
• Message ID: represents a unique message ID of 32 bits.
• Parameters: contains Type-Length-Format. Mandatory and optional LDP parameters are
also determined by this field.

50
6.1.3 LDP Type-Length-Value Format
The Type-Length-Value (TLV) format of LDP is presented in Table 6.3.

U F Type Length
Value
TLV Format

Table 6. 3 LDP Type-Length-Value Format

• U: represents unknown TLV bit.

• F: forwards unknown TLV bit. Used when U is set and the message needs to be
forwarded.
• Type: represents the encoding of the type of message.
• Length: represents the length of the Value field in octets.
• Value: variable length.

Overall, the LDP is responsible for explicit signaling, management, advertisement and
withdrawal of labels in the MPLS networking architecture.

6.2 LDP Messages

LDP messages are mainly categorized in four categories, as follows:

• Discovery Messages: used to inform about and maintain the presence of an LSR in the
network.
• Session Messages: used to establish, maintain and terminate sessions between LDP
peers.
• Advertisement Messages: used to create, change, and delete FEC label mappings.
• Notification Messages: used to provide advisory information and to signal error
information.

These messages are described in the following sections.

6.2.1 Notification Message

It is used by the LSR to let its peer know about important events. It may contain information on
one of the following

• Fatal error
• Processing outcome on an LDP message
• LDP Session state

51
6.2.2 Hello Message
During the Discovery mechanism, Hello messages are sent to let others know about LSR
presence.

6.2.3 Initialization Message

Initialization messages are transmitted during the LDP session establishment state.

6.2.4 KeepAlive Message

It is sent by an LSR to check the integrity of the LDP session transport session.

6.2.5 Address Message

It is sent by an LSR to its peer as an advertisement of its own interface addresses.

6.2.6 Address Withdraw Message

It is sent by an LSR to its peer as a withdrawal of its previously advertised interface addresses.

6.2.7 Label Mapping Message

It is sent by an LSR to its peer as an advertisement of valid FEC label bindings.

6.2.8 Label Request Message

It is sent by an LSR to its peer to request an FEC label binding or label mapping.

6.2.9 Label Abort Request Message

It is used to abort a previously sent label request message.

6.2.10 Label Withdraw Message

It is sent by an LSR to its peers in order for them to stop using the previously advertised specific
FEC label binding or mapping. As a result of this message, the FEC label mapping is removed.

6.2.11 Label Release Message

It is sent by an LSR to its peers when previously mapped FEC labels requested by the LSR or
advertised by its peers are no longer required.

6.3 Merging Labels

It is a technique in which multiple labels belonging to same FEC are combined to a single out-
going label in order to share network resources. This is performed only at the MPLS entry point
where FECs are assigned to specific packets. Decisions to forward or classify and packets

52
depend on the label and CoS field. If multiple labels have same the FEC, these can be easily
merged.

6.4 Label Distribution Modes

In the MPLS architecture, LSR makes the decision to bind a label to an FEC. Binding is performed
downstream with respect to that binding. When a label is assigned to a specific FEC, the
downstream LSR informs the upstream LSR. There are two types of downstream label
assignment and distribution.

1. Downstream on Demand Label Distribution

2. Unsolicited Downstream

The MPLS architecture allows LSRs to precisely request for specific FEC from its next hop. This is
mechanism is known as Downstream on Demand label distribution, i.e., when an LSR
specifically asks for a label binding. Figure 6.1 illustrates this procedure.

Figure 6. 1 Downstream on Demand Label Advertisements

In addition, the MPLS architecture enables LSRs to distribute label binding to other LSRs that
have not requested them. This process is known as Unsolicited Downstream label distribution

mode. It is the responsibility of the downstream LSR to advertise a label mapping to its
upstream LSRs. Figure 6.2 demonstrates the unsolicited downstream label advertisement
mechanism.

53
 Figure 6. 2 Unsolicited Downstream Label Advertisements

6.4.1 Independent Label Distribution Control

In this technique, an LSR can advertise its label mappings whenever it decides.

• In independent downstream on demand mode, an LSR immediately replies to label

mapping requests. It doesn’t wait for label mappings from its next hop.
• In independent downstream unsolicited mode, when an LSR has prepared an FEC, it may
advertise it to its adjacent neighbors.

As a result of the independent label distribution mode, a label can be advertised before
receiving downstream label information.

6.4.2 Ordered Label Distribution Control

In this mechanism, if an LSR has a FEC label mapping for its next hop, it can start advertising the
label mappings for that particular FEC. When the LSR is egress, it can start advertising label
mapping.

If there is no FEC next hop mapping or LSR is not egress, the LSR have to wait to receive
downstream label mappings and then map the FEC before transmitting it to its upstream LSR.

6.5 Label Retention

An LSR uses two modes for preserving the labels.

1. Liberal Label Retention Mode

2. Conservative Label Retention Mode

In the liberal retention mode, every label binding is obtained from a label distribution peer. LSR
decides whether the LSR is the next hop for the label binding. The LSR that preserved and
maintrained the received labels is an example of unsolicited downstream label advertisement
mode.
This mechanism is illustrated in Figure 6.3.

54
 Figure 6. 3 Liberal Label Retention Modes

The main advantage of this mode is that LSRs already have spare labels in their LIBs, which in
turn, results in faster responses to routing changes. Maintaining and distributing excessive large
label information is the main disadvantage of the liberal label retention mode.

6.6 Free Labels

Labels not bound by FECs are called free labels. There is a pool maintained by each LSR for such
labels.

At LSR initialization, all labels are kept in a pool. In order to determine supported label bindings,
the size of the pool is measured. If a new binding needs to be created, the LSR takes one label
from the pool of free labels and binds it with an outgoing packet. If a binding needs to be
disassociated, the LSR unbinds the label from the packet and returns it to the free label pool.

LSR is responsible for informing other LSRs of such events, i.e., whether a new binding is
created or an old one is removed. This mechanism is called remote label binding knowledge.

6.7 Label Space

For the assignment and distribution of labels the so-called label spaces are used. Label spaces
are categorized into two categories, as follows:

6.7.1 Per-Interface Label Space

It is a pool for interface-specific labels. This space typically uses interface resources for labels, as
follows:

• A FR interface that depends on Data Link Connection Identifiers (DLCIs) for assigning
labels.

55
 • Label controlled ATM interfaces depend on Virtual Channel Identifiers (VCIs) for label
assignment.

It is important to understand that when there are direct connections between LDP peers, the
per interface label space may be used. Consequently, these labels will only be used for traffic
transmitted through that interface.

6.7.2 Per-Platform Label Space

When interfaces are capable of sharing incoming labels, the per platform label space is used.

6.8 Label Stacking

MPLS has many great features; label stacking is an important feature best used in large MPLS
networks. A packet may be associated with many labels during its MPLS domain transition. The
ordering of these labels follows the Last-In-First-Out (LIFO) paradigm. When a supplementary
shim header is added to such packets, a so-called label stack is built. In the stack, the top most
label is currently used for forwarding the packet on the LSP. The label stack uses a stack bit,
indicating whether a label bit is first on the stack.

When a packet enters an MPLS network, a PUSH operation associates a label with the packet. If
the label needs to be removed, the label is removed through a POP operation. Many different
LSPs can be aggregated into a single LSP with the help of label stacking.

56
Figure 6. 4 Push and Pop Operations

57
 Chapter-7 Conclusions

7.1 Issues
MPLS is continuously expanding in backbone networks and is fulfilling an ever-growing range of
user demands and application requirements. There are nevertheless a few issues that impact
on the MPLS performance. These are discussed in the following sections.

7.1.1 Path Capacity and Multi-Class Traffic

58
Multi-Class traffic is highly influenced by real time applications QoS constraints, due to high
demands for available network resources. Such traffic has an impact on all the links in a
domain, because the traffic from real time applications concerns its delivery through any means
and utilizes every network resource regardless of dimensions. It is globally recognized that a
network has more physical paths than total links in it, which means the resources are not being
efficiently utilized.

It is recommended that there should be more links then paths, to increase the capacity of a
network. If, path capacity is enhanced it will result in improved network performance.

7.1.2 Load Dependent Parameters and Path Capacity

We know that the networks are combination of nodes and links. Router/switches are nodes and
transmission medium (wired/wireless) are links.

If we consider a network as a graph then it can be denoted by following

equation G=(N,L)

In the above equation, G denotes Network as a graph, N denotes nodes i.e. routers/switches
and L denotes all the links in that particular network. This graph helps in analyzing the net-
work’s capacity or path capacity.

The relationship between nodes and links is vital for an MPLS network. A network with more
nodes and links doesn’t guarantee performance efficiency, unless it is managed well according
to traffic load and demands. A network segment must not be over or under-utilized.

7.1.3 Load Dependent Parameters and Non-Linear Models

Almost all basic network models are shortest path setback. However, standard models can be
linear or nonlinear models.

A linear model is independent of load parameters such as link length or per unit flow cost.
Whereas, a nonlinear model is dependent of load parameters such as variance or mean of
delay.

Segments of IP network are usually linear network models and therefore nonlinear network
models have no role to offer. However, MPLS can take benefit of these nonlinear models.

7.2 Project Conclusions

59
The project Multi-protocol Label Switching Traffic Engineering with QoS confirmed the fact that
it is absolutely necessary to setup point-to-point LSPs in order to forward IP traffic through an
MPLS network. In this project report we concentrated on MPLS issues, QoS and TE in MPLS.
During our research, we examined the MPLS architecture and found out that MPLS is a
competent technique because it provides well-organized packet transmission, QoS, load
balancing, scalability, consistency, and end-to-end connectivity. MPLS is mainly a connection-
oriented architecture which easily integrates in existing IP networks.

In this report we used a hypothetical approach to MPLS TE and described every MPLS
component and concepts such as: LSR, LSP, LER, RSVP-T, RSVP, CR-LDP and labels. We have also
discussed the necessity and main advantages of using MPLS and TE and described how this is
implemented in MPLS. In order to analyze the performance of an MPLS network we provided
de-tails on this mechanism.

Finally, we concluded that the MPLS architecture can help reducing network congestion
together with an amazing possibility for added TE. This also provides an efficient way to utilize
all available network resources.

7.3 Future Work

MPLS has emerged as a powerful technology that forced several service providers to move their
services from ATM or Frame Relay networks to MPLS networks. For instance, Cisco is working
continuously to provide improved QoS in their network nodes and thus satisfy its users. IP high
speed links were the dominant backbone of the Internet, but MPLS is gradually taking control
and becoming an efficient replacement. This is mainly due to the fact that MPLS can be used in
combination with most routing protocols used in conventional IP networks allowing MPLS to
work in close collaboration with, and actually enhance the IP performance.

Applications such as VPN, VoIP, ERP, CRM, video conferencing and streaming need high QoS
requirements and these types of services are being well served in an MPLS environment.
Nevertheless, a progressive research work is required in order to make sure MPLS meets all
such requirements and fulfills these high demands.

60
References

rd
[1] B. A. Forouzan and S. C. Fegan, Data Communications and Networking, 3 Ed. New York,
United States: McGraw-Hill Higher Education, 2003.
th
[2] B. A. Forouzan and S. C. Fegan, Data Communications and Networking, 4 Ed. New York,
United States: McGraw-Hill Higher Education, 2006.
th
[3] A. S. Tanenbaum, Computer Networks, 4 Ed. New Jersey, United States: Prentice Hall, 2000.
th
[4] W. Stallings, Data and Computer Communications, 5 Ed. New Jersey, United States: Pren-
tice Hall, 2000.

[5] M. Sportack, IP Routing Fundamentals, United States: Cisco Press Publications, 1997.
th
[6] D. E. Comer, Internetworking with TCP/IP Principals, Protocols, and Architectures, 4 Ed. Vol.
1, New Jersey, United States: Prentice Hall, 2000.

[7] M. N. Aslam, Y. Aziz, Traffic Engineering with Multi-Protocol Label Switching, Blekinge
Institute of Technology, August 2008.

[8] International Engineering Consortium, White Papers: Multiprotocol Label Switching (MPLS),
International Engineering Consortium IEC, 2007 [Online, last accessed: April 17, 2010]:

http://www.iec.org/online/tutorials/mpls/index.html

[9] I. Ikram, Traffic Engineering with MPLS and QoS, Blekinge Institute of Technology, August
2009.

[10] G. H. Sabri, QoS in MPLS and IP Networks, Blekinge Institute of Technology, November,
2009.

[11] L. Lobo, MPLS Configuration on Cisco ISO Software, Cisco Press, ISBN 1-58705-199-0, Octo-
ber 21, 2005.

[12] A. Viswanathan, N. Feldman and Z. Wang, Evolution of Multiprotocol Label Switching,

Communication Magazine, IEEE 1998.

[13] Multi-Protocol Label Switching, Available: http://www.tdr.cesca.es/TESIS_UPC/AVAILABLE/TDX-

0323104-125205//Chapter1.pdf [Accessed: April. 17, 2010].

[14] Daniel O. Awduche, MPLS and Traffic Engineering in IP Networks, UUNET (MCI Worldcom)

61
[15] I. Hussain, Overview of MPLS Technology and Traffic Engineering Application, Internet
Technologies Division, Cisco Systems Inc. USA

[16] X. Xiao, A. Hannan, B. Bailey and L. M. Ni, Traffic Engineering with MPLS in the Internet,
GlobalCenter Inc., A Global Crossing Company, 141 Caspian Court, Sunnyvale, CA, Department
of Computer Science, 3115 Engineering Building, Michigan State University, East Lansing

[17] Juniper Networks, Technical Documentation: JUNOS 5.3 Internet Software Configuration
Guide: MPLS Applications, Juniper Networks Inc, 1999-2009. [Online, last accessed: May 12,
2010]:

http://www.juniper.net/techpubs/software/junos/junos53/swconfig53-mpls-apps/html/

[18] K. Kowalik and M. Collier, QoS Routing as a Tool of MPLS Traffic Engineering, Research In-
stitute for Networks and Communications Engineering (RINCE), Dublin City University, Ireland

[19] S. Gonzalez-Valenzuela and V. C. M. Leung, QoS Routing for MPLS Networks Employing
Mobile Agents, The University of British Columbia

[20] D. Caswell and S. Ramanathan, Using Service Models for Management of Internet Services,
IEEE Journal on Selected Areas in Communications, Vol. 18, No. 5, May 2000

[21] Internetworking Technology Handbook, Cisco Systems, Inc. [Online, last accessed: May 20,
2010]:

http://www.cisco.com/en/US/docs/internetworking/technology/handbook/QoS.html#wp1020
620

[22] L. Andersson, I. Minei, B. Thomas, LDP Specification, Network Working Group, RFC 5036,
Cisco Systesm, Inc. October 2007

[23] L. Fang, A. Atlas, F. Chiussi, K. Kompella, G. Swallow, LDP Failure Detection and Recovery,
AT&T, Avici Systems, Invento Networks, Juniper Networks, Cisco Systems, IEEE Communications
Magazine, October 2004

[24] A. Popescu, IP QoS: The Challenge of Building Efficient Networks, Blekinge Institute of
Technology. [Online, last accessed: June 16, 2010]:

https://bth.itslearning.com/Main.aspx?CourseID=1016

62
63