Topic

X Internet

QoS

4
OBJECTIVE
On completion of this topic, you should be able to:
1.

Discuss the concept of Internet QoS.

2.

Explain the operating principles of the Resource Reservation

Protocol (RSVP).

3.

Explain the concepts related to the Integrated Service (Intserv)

Model.

4.

Explain the concept related to the Differentiated Service (Diffserv)

Model.

5.

Compare and evaluate the advantages and disadvantages of

Diffserv and Intserv.

6.

Explain the operating principles of Multiprotocol Label Switching

(MPLS).

7.

Discuss the applications of MPLS to traffic engineering and virtual

private networks.

8.

Describe the key features of Internet Protocol Version 6 (IPv6).

9.

Compare and evaluate the advantages and disadvantages of IPv6

and IPv4.

X OVERVIEW
This topic introduces you to several technologies that provide Internet QoS
that is, Internet quality of service: Resource Reservation Protocol (RSVP),
Integrated Service (Intserv) model, Differentiated Service (Diffserv) model,
Multi-Protocol Label Switching (MPLS) and Internet Protocol Version 6 (IPv6).

TOPIC 4 INTERNET QOS W 81

Well begin with the basic concept of RSVP; youll learn how RSVP can be used
to support the Intserv model, and the limitations of the resulting Intserv/RSVP
model will be discussed. Afterwards, youll be introduced to the Diffserv model,
which can ease some of these problems, such as scalability.
The operating principles of MPLS are explained next. Two important MPLS
applications traffic engineering and virtual private networks are covered.
Finally, you are introduced to the key features of the next generation Internet
Protocol: IPv6.
Good luck as you begin your study!

INTRODUCTION

In the previous topics, you were introduced to several advanced technologies

that support high-speed connections between users and service providers. For
example, an ATM-over-ADSL access network can offer a cost-effective solution
to provide broadband services to the mass market. Such high-speed networks
form the physical infrastructure that supports bandwidth-demanding
applications, such as video-on-demand and interactive games, over the Internet.
Although the capability to transmit information as quickly as possible is
obviously desirable, it is not completely satisfactory from the users perspective.
The existing Internet can only provide best-effort service, i.e. it tries its best to
forward user traffic. Using existing mechanisms, however, there can be no
guarantees regarding loss rate or delays because packets can be dropped
indiscriminately during congestion. While this best-effort service works fine for
most traditional Internet applications such as email, it is totally unacceptable for
emerging real-time multimedia applications such as Internet telephony or videoon-demand (VOD). Random packet-dropping can seriously degrade the audible
or visual quality of a streamed movie. In addition, some premium customers are
willing to pay higher service charges to obtain better service (e.g., lower delay
latency, higher data rate). Thus, there is growing demand to enhance the existing
Internet to provide better transmission services i.e. exceeding best-effort. This
leads us to the study of Internet Quality-of-Service (QoS).
This topic introduces you to several service models and mechanisms proposed
by the Internet Engineering Task Force (IETF) to meet the demand for Internet
QoS in the Next Generation Internet. The Resource Reservation Protocol (RSVP),
Integrated Service (Intserv) model, Differentiated Service (Diffserv) model,
Multi-Protocol Label Switching (MPLS) and Internet Protocol Version 6 (IPv6)
will all therefore be covered.

82

X TOPIC 4 INTERNET QOS

These efforts are all emerging technologies to provide Internet QoS at different
levels in the OSI model. Intserv and Diffserv represent new service models at the
transport layer to extend the current best effort service to accommodate new
applications such as real-time video. IPv6 is an enhanced version of Internet
protocol designed to fix some of the problems of the current IP (e.g. the potential
address shortage problem). RSVP is a new signalling protocol for setting up
paths and reserving resources to support both Intserv and Diffserv. MPLS is a
connection scheme between the link layer and network layer (known as layer 2.5)
to forward packets using simplified address labels. By applying these
technologies together, we can build a framework to address future Internet QoS
requirements.
The remaining parts of this topic describe the general principles underlying these
models and protocols, and the ways they can be interconnected to support QoS.
Before we go into details of individual topics, however, we shall pause to take a
look at the issue of QoS in the next section. In particular, we shall clarify the exact
meaning of QoS.

WHAT IS QOS?
The term Quality of Service (QoS) is not new to you. In other earlier topic, you
learned that one distinct advantage of ATM is its ability to guarantee QoS. Well,
what does QoS really mean?
We can begin to define QoS by pointing out that it is a set of service requirements
to be met by the network while transporting a packet stream from a source to a
destination (unicast or multicast). The common QoS parameters used in Internet
services are:


bandwidth: the average usable and available bandwidth over the link at any
time;

delay: the average end-to-end delay caused at network level at any time;

delay jitter: the average difference of the various delay times over the link;

packet loss probability: the average probability of packet loss over the link
over a length of time.

As you know, different applications have different requirements for bandwidth,

delay, and packet loss probability. Therefore, QoS must also be a measurable
level of service delivered to network users. Such QoS can be provided by
network service providers in terms of an agreement, known as a Service Level
Agreement (SLA) made between network users and service providers. Recall
that, for example, that the peak rate or cell loss ratio is specified in the traffic

TOPIC 4 INTERNET QOS W 83

descriptors in ATM. Similarly, a user may specify an SLA with a minimum

bandwidth of 1 Mbps for transmission. Hence QoS features provide the ability to
manage traffic intelligently across a network.

Some applications require bandwidth and delay guarantees, referred to as

quantitative applications, while others are more qualitative. For example, voice
applications have stringent delay requirements and can tolerate minimal packet
loss. Conversely, an FTP file transfer may be insensitive to delays but very
sensitive to packet drops. To accommodate these differences, QoS generally
assigns traffic flows to one of two categories:


Guaranteed service (quantitative) reserves a designated amount of

bandwidth from end to end and can guarantee a specified delay tolerance for
the exclusive use of an application or even aggregated sessions. This type of
service is best for applications that require a specific amount of bandwidth
with long duration. For example, a video conference call may require at least
64 kbps for 30 minutes.

Differentiated service (qualitative) provides simple prioritization.

Applications are detected at the ingress and assigned SLAs. It is the SLAs that
drive which QoS mechanisms the router will use, for example, which queue
will be used to place traffic, and, if congestion requires packets to be
dropped, which drop priority will be designated.

After completing our overview of these introductory QoS concepts, we shall start
our discussion on individual technology to support QoS. First, however, please
read the following paper by Xiao, which presents an overview of different
technologies we are going to cover in this topic. Dont worry if you do not fully
understand the technical details of each individual topic. This topic provides
detailed discussion of each topic later. However, this reading gives you a global
picture of various technologies and the relationships between them. As you work
through this topic, you will find it helpful to revisit this paper from time to time.

84

X TOPIC 4 INTERNET QOS

READING 4.1
Xiao, X and Ni, L M Internet QoS: a big picture,
http://www1.cs.columbia.edu/~zwb/my/oral/qos/netmag/qos.pdf
This article provides a framework for the emerging Internet QoS.
Please read the paper up to the section on Intserv and RSVP. Very
good examples are provided in individual topics and we shall reread
this paper later.

The next section introduces Resource Reservation Protocol (RSVP). First,

however, complete Self-test 4.1.

SELF-TEST 4.1
1.

How would you describe the most common problems related

to service delivery on todays Internet?

2.

List and briefly describe three QoS parameters commonly used

in ATM networks.

3.

Explain how ATM networks provide a QoS guarantee. Do you

think this can be used on the Internet?

RESOURCE RESERVATION PROTOCOL (RSVP)

Although it evolved from a pure data network, the Internet currently carries
different types of traffic with diverse characteristics and service requirements.
Both real-time and non-real-time traffic is sent through the networks using an
Internet Protocol (IP). The IP bearer service provides no guarantee of service
quality in terms of error, delay, or bandwidth. The network will attempt to do the
best it can to deliver the message, so, as youve seen, such service is called besteffort service. If there is sufficient bandwidth in the network, then all of these
service requirements will be met. Nevertheless, packets will be dropped
indiscriminately during congestion. This may not be a problem for traditional
data traffic because the Transport Control Protocol (TCP), which operates on top

TOPIC 4 INTERNET QOS W 85

of the IP, can be used to provide reliable communication by retransmitting

missing packets.
TCP is not suitable, however, for delay-sensitive applications such as audio or
video. Alternatively, the User Data Protocol (UDP) is used to transport real-time
multimedia service. Since UDP does not guarantee delivery of packets, audible or
visual quality will be severely affected by dropped packets. For instance, Internet
streaming technologies such as Real Audio/Video use UDP as their transporting
protocol. (You can verify this by looking at the software.) If you watch CNN with
Real Video over the Internet during busy times, you will most likely see low
quality video playback because a lot of packets are being lost.
To guarantee service quality for a given type of traffic, separate resources can be
reserved. For instance, you can reserve 32 kbps of dedicated bandwidth for
transmitting voice packets. Using this basic concept, the Resource Reservation
Protocol (RSVP) is a signalling protocol that can be used by hosts to request
resource reservation through a network. A sender host uses an RSVP PATH
message to declare the bandwidth for data flow. The PATH message traverses
several routers to its destination. The receiving host uses the RSVP RESV
message to reserve an amount of bandwidth. The RESV message traces the path
back to the sender host, reserving the resources along the intermediate routers.
Routers may accept or reject the reservation request, based on available
resources. Sophisticated built-in multicast support also merges the reservations
for the same flow coming from different receivers along the intermediate routers.
You should note that RSVP is based on a soft state mechanism: reservations
expire after a time-out period unless refreshed.
Keep in mind that data are not really transmitted during the reservation process.
Instead, RSVP can ensure the reserved resources will be available if and when
data are actually sent through the network.
Youve now got two readings to work through that give you more detail
regarding RSVP. Please read Section 6.8 in Kuroses book for a detailed
description of RSVP. The examples in this article can help you understand the
operation principle of RSVP. The review paper by Liu also briefly overviews
RSVP, and you can go through it quickly.

86

X TOPIC 4 INTERNET QOS

READING 4.2
Kurose, J and Ross, K (2001) Section 6.8, RSVP, Computer
Networking: A Top-Down Approach Featuring the Internet,
Addison-Wesley, pp 53949.
This article provides a brief review of the different emerging high
speed access technologies for home users. It gives you a general idea
of different technologies before you study them in depth in later
parts of this topic.

READING 4.3
Liu, C Multimedia Over IP: RSVP, RTP, RTCP, RTSP
<http://www.et.fnt.hvu.nl/docenten%5Clab/project_telematica_DT/
2000_voorjaar/project_16_streaming_DT/ip_multimedia.pdf>
This article provides a brief survey of the different protocols to
support multimedia over the Internet. You can just read the part on
RSVP.

SELF-TEST 4.2
1.

What are the two types of message in RSVP used for

reservations? Please describe their functions.

2.

Who is responsible for initiating a reservation request in RSVP?

3.

What are the necessary procedures that must run on each node
to set up a reservation?

4.

What is the meaning of soft state as related to RSVP?

As youve now learned, RSVP is only a protocol for reserving resources. It does
not specify how the network provides the reserved bandwidth to the data flow to
guarantee QoS. In other words, RSVP alone will not automatically provide the
required QoS, so it must work with other protocols to meet service requirements.
The Integrated Service (Intserv) model proposed by IETF is a framework that

TOPIC 4 INTERNET QOS W 87

provides service classes in addition to the traditional best-effort services. We

shall discuss this model in the next section.

INTEGRATED SERVICES (INTSERV)

The Integrated Services model is a proposed extension to current Internet
architecture and protocols that is intended to provide integrated services that
support both real-time and current non-real-time IP service. The basic concept is
to foresee a set of service models to be provided on the Internet, moving beyond
the best-effort model. Two such service models for real-time applications are
defined:
1.

Guaranteed service provides strict boundaries for the delay and loss
probabilities for packets from a given flow, provided that the flow complies
with a traffic contract.

2.

Controlled-loaded service provides a QoS comparable to best-effort service

under unloaded conditions.

The implementation of the Intserv model includes mechanisms for resource

reservation, admission control and queue management. Applications requiring
guaranteed or controlled-load service must set up the path and reserve resources
before transmitting their data. The call admission control mechanism determines
whether a request for resources can then be accepted. A queue management
mechanism is used to schedule packet transmissions according to the QoS
requirement. Thus, the Intserv model requires routers along the route to reserve
adequate resources and maintain the state of flow.
In the previous section, you learned that RSVP is a network control protocol
allowing data receivers to request a special end-to-end quality of service for their
data flows. The Intserv model can use RSVP to reserve necessary resources in
routers along a path before transmission commences. Thus RSVP is integrated as
a main component of Intserv.
Please read the following paper by Iyer for a detailed introduction to the Intserv
model. You should pay particular attention to the section on classes of service.
Section 6.7 in Kuroses book also briefly describes the Intserv model, and you can
go through it quickly.

88

X TOPIC 4 INTERNET QOS

READING 4.4
Iyer, J. Integrated Services on the Internet, <ftp://ftp.netlab.ohiostate.edu/pub/jain/courses/cis788-97/integrated_services/ index.htm >
This article provides a good overview of the Integrated Service model.
The discussion of the two service classes is clear and easy to
understand.

READING 4.5
Kurose, J and Ross, K (2001) Section 6.7, Integrated Service, Computer

Networking: A Top-Down Approach Featuring the Internet,

Addison-Wesley, pp 53639.
This article provides a brief review of the Integrated Service model.

The Intserv/RSVP architecture is able to provide QoS guarantees to individual

flows through reservation. Nevertheless, the Intserv model has scalability
problems because it requires maintenance of per-flow state on the network. This
restricts the wide deployment of the Intserv model as a solution for Internet QoS.
Consequently, this has led to the development of a more scalable and flexible
model known as Differentiated Service (Diffserv), which will be discussed in the
next section.

SELF-TEST 4.3
1.

What are the new classes of service being added to the Intserv
model? What are the two parameters used to specify traffic
characteristics and QoS requirements?

2.

Describe the components required to implement the Intserv.

3.

List some of the disadvantages of Intserv.

TOPIC 4 INTERNET QOS W 89

ACTIVITY 4.1
Suppose you are going to set up a multi-party video conference
using Intserv. Please explain the necessary procedures.
Your tutor will discuss with you the approach you might take
during your next tutorial.

DIFFERENTIATED SERVICES (DIFFSERV)

Due to the difficulties related to implementing the Intserv model, the IETF
recently proposed the Differentiated service (Diffserv) model, which enables
either end-to-end or intradomain service discrimination. The Diffserv model can
specify and control network traffic by class in such a way that certain types of
traffic get priority. For instance, voice traffic, which requires a relatively
uninterrupted flow of data, might get priority over other kinds of traffic. In
contrast to Intserv, user flows in Diffserv are only controlled at the edge of the
network and then aggregated into a small set of traffic classes.
In the Diffserv model, traffic is classified and possibly conditioned while entering
a network. Based on the result of this classification, classes of traffic are then
attributed to different behaviour aggregates. Each of these traffic classes is
identified by a field, called a Diffserv Codepoint (DSCP), in the IP header. Within
a Diffserv network, packets are forwarded according to the specific queuing
behaviour (known as per-hop behaviour or PHB) associated with their DSCP.
Per-hop behaviour defines how an individual router treats an individual packet
when sending it over the next hop through the network. In addition, traffic
classes from many flows having similar QoS requirements are marked with the
same DSCP. These flows can then be aggregated to a common queue. Since the
DS field is set at the network boundaries, no per-flow state is necessary at the
network core. This eliminates the scalability problem characteristic of the Intserv
model.
By establishing a way of delivering differentiated per-hop forwarding behaviour
to IP packets, Diffserv allows a shared network to accommodate different QoS
levels for traffic streams using the same infrastructure. Diffserv enables ISPs to
define classes of service (CoS) to support particular traffic requirements and to
offer premium services for special data types such as voice. In other words, we
can define gold, silver or bronze services for different users. The standard
will also specify components that might be used in usage-based pricing, helping
ISPs balance customer demand with available bandwidth.

90

X TOPIC 4 INTERNET QOS

In the readings that follow, youll begin by reviewing the section on Diffserv
from the Xiao paper you overviewed earlier in this topic. Basic Diffserv concepts
are covered very well in this paper. You should also read Section 6.9 in Kuroses
book for a detailed description of the Diffserv model. These two readings do a
good job describing the Diffserv models fundamental concepts and operating
principles.
Most RSVP applications are associated with the Intserv model. In Bernets paper,
however, the application of RSVP is extended to Diffserv. In particular, he
proposes the construction of a QoS network combining RSVP and Diffserv that
realizes the advantages of each.
Since Diffserv is still in the drafting stage, you can find the most updated
information about it on a website from the Diffserv Working Group of IETF
(<http://www.ietf.org/html.charters/diffserv-charter.html>). You can also
listen to a virtual seminar by Prof. Jain on the issues related to providing QoS on
data networks if you choose.

READING 4.6
Kurose, J and Ross, K (2001) Section 6.9, Differentiated Service,

Computer Networking: A Top-Down Approach Featuring the

Internet, Addison-Wesley, pp 54956.
This article provides a brief review of the Diffserv model.

READING 4.7
Bernet, Y (2000) The complementary roles of RSVP and
differentiated services in the full-service QoS network, IEEE
Communication Magazine, 38(2), Feb: 15462.
The paper proposes a model combining RSVP with Diffserv to
enable QoS in a scalable manner.
(optional)

TOPIC 4 INTERNET QOS W 91

READING 4.8
Carpenter, B and Kandlur, D (1999) Diversifying Internet delivery,
IEEE Spectrum, 36(11), Nov: 5761.
This paper provides some insight on the issue of differentiated
service; you can just scan through it. It is available from the OUHK
library or over the Internet.

READING 4.9
Jain, R. Quality of Service in Data Networks, The Ohio State
University, CIS788.08Q Class Lecture.
<http://www.cis.ohio-state.edu/~jain/cis788-99/h_6qos.htm>

This virtual lecture lasts for about 1.5 hours. Various techniques including
ATM, Intserv and Diffserv models to provide QoS in data networks are
overviewed. You can listen up to the part on the problems on Diffserv. You can
return to the part on MPLS after you have finished the next section of this topic.
Both Intserv and Diffserv are new service models aimed at providing Internet
QoS. There have also been other recent efforts by the IETF to enhance the speed,
scalability, and service provisioning capabilities on the Internet. One such rapidly
emerging standard is MultiProtocol Label Switching (MPLS). We shall discuss
the MPLS architecture in the next section. First, however, you should complete
Self-test 4.4 and Activity 4.2.

SELF-TEST 4.4
1.

What approaches characteristic of the Diffserv model solve some

of the Intserv models problems?

2.

What is the function of PHB? What are the four PHBs provided by
Diffserv?

3.

Explain the decision to be made in a customer domain given a

service level agreement.

92

X TOPIC 4 INTERNET QOS

ACTIVITY 4.2
Suppose you are going to set up a video conference using Diffserv.
Please explain the necessary procedures required.
Your tutor will discuss with you the approach you might take
during your next tutorial.

MULTIPROTOCOL LABEL SWITCHING (MPLS)

In recent years, the computing/telecommunications industry has been searching
for an approach to combine the desirable features of IP and ATM. We would like
to combine the scalability and flexibility of IP routing with the performance, QoS
and traffic management ATM switching can provide. In earlier topic, you were
introduced to the IP-over-ATM model, which allows IP and ATM to work
together without changes to either protocol. Nevertheless, this overlay network
approach does not take full advantage of ATM, and is inefficient in certain
aspects.
MultiProtocol Label Switching (MPLS) introduces a new forwarding paradigm
for IP networks. The idea is similar to that in ATM: a path is first established
using a signalling protocol. Then a label in the packet header instead of the IP
destination address is used for making forwarding decisions in the network.
As a result, MPLS can introduce the notion of connection-oriented forwarding
over an IP network. MPLS is called multiprotocol because it works with the IP,
ATM, and frame relay network protocols. With reference to the OSI model,
MPLS allows most packets to be forwarded at the layer 2 (switching) level rather
than at the layer 3 (routing) level.
The key architectural principle of MPLS is a clean separation of an IP routers
function into two parts: forwarding and control. This separation enables effective
service integration, including QoS transparency, between service layers. The
forwarding function is responsible for how data packets are transmitted between
IP routers. Label swapping is used, similar to ATM switching using VPI or VCI.
A router capable of MPLS is therefore a label switching router (LSR).
MPLSs control function consists of network layer routing protocols that
distribute routing information between LSRs, and label binding procedures for
converting this routing information into the forwarding tables needed for label
switching.

TOPIC 4 INTERNET QOS W 93

Please read the tutorial on MPLS by Web ProForum, which provides a very good
overview of the MPLS technology. The basic principles of MPLS are also covered
in Xiaos paper, and the examples there are illustrative.
With MPLS a virtual connection is established between two points on a pure
datagram network, and the connection in turn carries datagram traffic. MPLS can
provide many of the same advantages of a connection-orientated network while
still retaining the underlying efficiency and operation of a datagram network. In
particular, the most important benefit of MPLS is the ability to perform traffic
engineering.
Traffic engineering is a loose term that encompasses many aspects of network
performance. These include the provision of a guaranteed QoS, improving the
utilization of network resources by spreading traffic evenly across the network,
and providing features for quick recovery when a node or link fails. The
applications of MPLS to traffic engineering in IP networks are discussed in
Awduches paper in the readings that follow.
Another important application of MPLS is its ability to support virtual private
networks (VPNs). Such networks are useful in providing internal networks to
distributed organizations. A typical example is the interconnection of several
regional offices with a corporate headquarters. Basic VPN concepts are discussed
in Dicksons paper.
Finally, you can choose to listen to another optional online lecture by Prof. Jain
on MPLS, which lasts about 35 minutes.

READING 4.10
The International Engineering Consortium, Multiprotocol label
(MPLS)
Tutorial,
Web
ProForum
Tutorials
<http://www.webproforum.com/mpls/index.html>

switching

This tutorial provides an in-depth look at the technology behind

MPLS, with an emphasis on the protocols involved. The tutorial also
discusses why MPLS is an important component in the deployment of
converged networks.

94

X TOPIC 4 INTERNET QOS

READING 4.1
Xiao, X and Ni, L M Internet QoS: a big picture,
<http://www1.cs.columbia.edu/~zwb/my/oral/qos/netmag/qos.pdf>
Please read the section on MPLS and follow the examples carefully.
In particular, pay close attention to how MPLS can be used together
with differentiated services to provide QoS.

READING 4.11
Awduche, D (1999) MPLS and traffic engineering in IP networks,
IEEE Communications Magazine, 37(12), Dec: 4247.
The article discusses the application of MPLS to traffic engineering in
IP networks.

READING 4.12
Dickson, K (1999) The ABCs of VPNs a virtual private networking
primer for enterprises, Cisco Packet Magazine, 4th Quarter 1999.
<http://cisco.com/warp/public/784/packet/oct99/vpn.html>
This paper discusses the basic concept related to VPNs.
(Optional)

TOPIC 4 INTERNET QOS W 95

READING 4.13
Davie, B (1999) Multiprotocol label switching service providers to
benefit from new functionality, Cisco Packet Magazine, 2nd
Quarter 1999.
<http://www.cisco.com/warp/public/784/packet/apr99/6.html>
This article explains how MPLS provides a variety of benefits,
including support for virtual private network (VPN) services, traffic
engineering, and IP +ATM integration.

READING 4.14
Jain, R,. IP Switching and Label Switching, The Ohio State University,
CIS788.08Q Class Lecture.
http://www.cis.ohio-state.edu/~jain/cis788-99/h_4ipsw.htm
The whole lecture lasts for about an hour, but you can focus your
study on the MPLS part starting at around the 25 minute mark.

In this topic, we have so far discussed several emerging service models and
techniques to provide Internet QoS. Nevertheless, the existing Internet protocol
known as Internet Protocol Version 4 (IPv4), which was developed in early 80s,
may not be able to accommodate future service needs and traffic volume growth.
This has led to the development of next generation Internet Protocol known as
IPv6, which will be our next topic. As usual, however, complete the following
self-test before you go on to the next section.

96

X TOPIC 4 INTERNET QOS

SELF-TEST 4.5
1.

Explain the main idea behind MPLS.

2.

What modules does a MPLS router usually include?

3.

What are the two signalling protocols defined in MPLS?

Identify a difference between them.

4.

Explain the advantages of applying traffic engineering to the

Internet.

INTERNET PROTOCOL VERSION 6 (IPV6)

Growth is the basic issue behind the need for a next generation Internet Protocol.
The current Internet protocol IP Version 4 (IPv4) mainly serves the
computing market, which has been the driving force of the growth of the Internet
for the past 15 years. The next growth phase, however, will probably be driven
by other markets in addition to the computing market. For instance, we have
witnessed more and more real-time applications such as Internet Telephony
being deployed in recent years. Despite the great variations in traffic behaviour
over the years, very few changes have been made in IPv4. IPv4 has therefore
gradually developed problems handling the continuous growth of users and
applications. In particular, there is a growing shortage of IPv4 addresses (they
are only 32 bits long), which are needed by all new machines added to the
Internet.
Internet Protocol Version 6 (IPv6) is the next generation protocol designed by
the IETF to replace the current IPv4. IPv6 has also been called IPng (IP Next
Generation). IPv6 aims to fix IPv4s main problems (e.g. limited IP address) and
to add improvements such as routing and network autoconfiguration. The most
important changes are as follows.


Expanded routing and addressing capabilities: IP address size increases from

32 bits to 128 bits and a new type of address called an anycast address is
introduced.

Header format simplification: some IPv4 header fields have been dropped or
made optional to streamline the header size to 40 bytes.

Quality-of-Service capabilities: flow label and priority field are added.

There will again be several readings you need to work through. First, please read
Section 4.7 in Kurose and Rosss book for the general information about IPv6.

TOPIC 4 INTERNET QOS W 97

This reading also contains useful links to other references related to IPv6. The
paper by Hinden provides a detailed overview of IPv6 but it is rather old (1995).
Thus, you can just quickly scan through it. McNealis paper evaluates the relative
merits of IPv6 as compared with IPv4.

READING 4.15
Kurose, J and Ross, K (2001) Section 4.7, IPv6, Computer Networking:
A Top-Down Approach Featuring the Internet, Addison-Wesley, pp
34148.
This article introduces the general ideas behind IPv6.

READING 4.16
Hinden, R IP Next Generation Overview.
http://playground.sun.com/pub/ipng/html/INET-IPng-Paper. html
This paper provides a good overview of the IP Next Generation (i.e.
IPv6). The proposed IPv6 functions are covered. Since the paper was
written in 1995, however, some of the contents may be outdated.
(Optional)

READING 4.17
Reading 4.17: McNealis, M (1998) IP crossroads, migrate to IPv6 or
evolve with IPv4?, Cisco Packet Magazine, 3rd Quarter 1998.
<http://www.cisco.com/warp/public/784/packet/july98/9.html>
This article compares the advantages of IPv6 and IPv4.

98

X TOPIC 4 INTERNET QOS

READING 4.18
Reading 4.18: IP Next Generation (IPng) <http://playground.sun.
com/pub/ipng/html/ipng-main.html>
This website provides an excellent online source of information about
IPv6.

READING 4.19
Answers to Frequently Asked Questions about IPv6.
<http://www.wide.ad.jp/wg/finish/014_ipv6.html>

IPv6 was designed to make an evolutionary step beyond IPv4. It was not,
however, a design goal to take a radical step away from IPv4, in order to ensure
compatibility. Functions which work in IPv4 are retained in IPv6, while functions
which didnt work have been removed. IPv6 is expected to gradually replace
IPv4, with the two co-existing for a couple of years during the transition period.

SELF-TEST 4.6
1.

What is the major cause driving the development of IPv6?

2.

Identify an application of unicasting.

3.

What is the function of Label Flow in IPv6?

4.

Which functions are removed from IPv6? Explain why these

features are advantageous to high-speed networking.

5.

Despite the advanced features added to IPv6, what major

obstacles stand in the way of IPv6 replacing the existing Internet
protocol?

TOPIC 4 INTERNET QOS W 99

We have discussed several emerging networking technologies. You have been

introduced to Gigabit Ethernet and ATM, which are likely to be the dominant
backbone networks in LANs and WANs, respectively. These high-speed
backbones provide the necessary bandwidth to build the infrastructure for the
future information highway.
In the previous topic, you learned about three residential broadband access
techniques: ADSL, Cable Modems and Fixed Wireless Networks. These new
access mechanisms enable SOHO users to run bandwidth-greedy applications
such as video-on-demand at home. With this increased capacity, however, users
are no longer satisfied with the current best effort service delivered by TCP/IP;
users want a better quality of service. This has motivated the industry to develop
new protocols to enhance the existing Internet to better be able to provide QoS. In
this topic, we have discussed several technologies to provide Internet QoS.
We started our discussion by introducing the requirements for service quality
and the concept of QoS. You were then introduced to RSVP, a signalling protocol
to all data receivers that reserves an end-to-end QoS for its data flows. The
operating principles behind RSVP were then covered. Next, we discussed how
RSVP can be used in the Integrated Service model, which can support both besteffort and real-time services. Nevertheless, the Intserv/RSVP model must
maintain state information for each traffic flow, which limits potential network size.
Because of the difficulties implementing and deploying the Intserv/RSVP,
Diffserv has been introduced. Diffserv is aimed at providing scalable and flexible
service differentiation. Based on per-hop behaviour, Diffserv routers can support
multiple service classes. As Diffserv is still in the development stage, new
features and functions may be added to the standard later.
MPLS is a new protocol standard intended to speed up network traffic flow and
simplify management. MPLS can support multiple protocols such as IP, ATM etc.
MPLS enables us to do traffic engineering on the Internet. In addition, MPLS
provides a cost-effective paradigm for setting up virtual private networks.
Finally, the key features of the next generation Internet Protocol IPv6 were
overviewed, and we discussed the advantages of IPv6 over existing IPs.
In short, providing QoS to the Internet is a complicated issue that requires new
protocols to cooperate with each other. The technologies weve discussed are all
emerging efforts in an Internet community pushing for better service.

100 X TOPIC 4 INTERNET QOS

To wrap up, you can look at the hierarchy of different network layers in Figure
4.1 on the next page. This figure illustrates the interconnections between the
different technologies already covered in this module. That is, the ATM over
ADSL platform provides the necessary broadband access for users. MPLS enables
simple integration of ATM and IP technologies. The Intserv and Diffserv models
with RSVP provide QoS to Internet users. Figure 4.1 therefore shows a possible
structure for building a future broadband-service Internet.

Figure 4.1: A possible hierarchy for future broadband-service Internet

In addition to the recent explosion of Internet traffic volume, we have also

witnessed tremendous growth in wireless communications. For example, the
number of mobile phone users in Malaysia has increased tremendously over the
past three years. In addition, there is growing demand for data services (in
addition to voice) in mobile communications. With the introduction of advanced
technologies such as WAP and Bluetooth, we shall see more non-voice
applications being added. Furthermore, a so-called Third Generation (3G)
system will emerge to provide a new communication infrastructure to support
high-speed data service for mobile users. This 3G system will enable users to
enjoy broadband services on the road.

TOPIC 4 INTERNET QOS W 101

Diffserv

Differentiated Service

DSCP

Diffserv Codepoint

IETF

Internet Engineering Task Force

Intserv

Integrated Service

IPv6

Internet Protocol Version 6

LSR

Label Switching Router

MPLS

Multi-Protocol Label Switching

PHB

Per-Hop Behaviour

RSVP

Resource Reservation Protocol

SLA

Service Level Agreement

TCP

Transport Control Protocol

UDP

User Data Protocol

VPN

Virtual Private Network

SUGGESTED ANSWERS TO SELF-TEST QUESTIONS

Self-test 4.1
1.

Todays Internet can only provide best-effort service, which means it will
try its best to forward user traffic, but can provide no guarantees regarding
loss rate, bandwidth, delay, delay jitter, etc. However, new applications
such as voice transmission require better transmission services than besteffort.

2.

The common QoS parameters are: Cell Loss Ratio, Cell Delay Variation,
Peak Cell Rate, etc.

3.

ATM exercises call admission control. That is, when a network accepts a
connection request, the user and the network agree on a traffic contract for
the duration of the connection. With this contract established, the network
guarantees the requested service demand for the connection as long as the
source traffic complies with specified limits. If the network does not have

102 X TOPIC 4 INTERNET QOS

sufficient resources to accommodate the connection request, it simply

rejects the call.
Yes, this idea can be applied to the Internet provided the Internet protocol
is able to reserve resources on request and exercise call admission control.

Self-test 4.2
1.

The two types of message: PATH and RESV.

PATH messages are sent periodically from the sender to the multicast
address. A PATH message contains flow specs to describe the sender
template (data format, source address, source port) and traffic
characteristics. This information is used by receivers to find the reverse
path to the sender and to determine what resources should be reserved.
Receivers must join the multicast group in order to receive PATH
messages.
RESV messages are generated by receivers and contain reservation
parameters including flow specs and filter specs. The filter spec defines
what packets in the flow should be used by the packet classifier. The flow
spec is used by the packet scheduler, and its content depends on the
service. RESV messages follow the exact reverse path of PATH messages,
setting up reservations for one or more senders at every node.

2.

The reservation requests are initiated by the receivers.

3.

(a)

Policy control determines whether the user has administrative

permission to make a reservation.

(b)

Admission control keeps track of the system resources and determines

whether the node has sufficient resources to supply the requested
QoS.

4.

In RSVP, reservations for bandwidth (state information) are associated with

a timer. A reservation can only be maintained if it is refreshed periodically.
Otherwise, the reservation will be removed after the time out.

Self-test 4.3
1.

Guaranteed Service Class

Controlled Load Service Class

TOPIC 4 INTERNET QOS W 103

2.

3.

The four components are:



signalling protocol (e.g. RSVP)

admission control routine

classifier

 packet scheduler.
The disadvantages of Intserv:


New software or firmware must be added to all routers along the

network path. The demands on routers are high.

The amount of state information increases proportionally with the

number of flows. This leads to scalability problems.

Self-test 4.4
1.

In Diffserv, only simple functions are placed within the network core, and
more complex control operations are implemented towards the edge of
the network.

2.

PHB determines the forwarding rate of a particular type of traffic according

to the class marked on the packets in that traffic.
The four PHBs are:

3.

default behaviour

class-selector behaviour

expedited forwarding behaviour

assured forwarding behaviour.

For a given SLA, a customer domain must decide how its host shares the
services shared by the SLA. This process is known as service allocation.

Self-test 4.5
1.

MPLS involves setting up a specific path for a given sequence of packets,

identified by a label put in each packet. This therefore saves the time
needed for a router to look up the address for the next node to forward the
packet to. MPLS integrates the label switching forwarding paradigm with
network layer routing.

104 X TOPIC 4 INTERNET QOS

2.

3.

An MPLS router might include:



a unicast routing module, which builds up the routing table using

conventional IP routing protocols (OSPF, Enhanced IGRP, and so forth),
assigns labels to the routes, and distributes labels using the Label
Distribution Protocol (LDP);

a traffic engineering module, which enables explicitly specified labelswitched paths to be set up through a network for traffic engineering
purposes;

a VPN module, which builds VPN-specific routing tables using the

Border Gateway Protocol (BGP) and distributes labels corresponding to
VPN routes.

Resource Reservation Protocol extension

Constraint Routed Label Distribution Protocol (CR-LDP)
RSVP uses the refreshed-soft-state model while CR-LDP is a hard-state
protocol.

4.

Using traffic engineering, we can arrange the traffic flow through the
network such that congestion caused by uneven network utilization can be
avoided. This can improve the throughput and delay response in the
network.

Self-test 4.6
1.

There is a growing shortage of IPv4 addresses, which are needed by all new
machines added to the Internet. IP addresses are lengthened from 32 bits to
128 bits in IPv6.

2.

A packet sent to an anycast address will be delivered to any one of a group

of hosts. Applications try to download a MP3 file from a set of mirror sites.

3.

Packets can be identified as belonging to a particular flow so that packets

that are part of a multimedia presentation that needs to arrive in real time
can be provided with a higher quality-of-service relative to other
customers.

4.

The following functions are removed from IPv6:



Fragmentation/Reassembly

Checksum

Options

TOPIC 4 INTERNET QOS W 105

In high-speed networks, we wish to transfer packets as quickly as possible.

Removing these functions can reduce the processing required in each
router.
5.

The major obstacles related to IPv6 are:



It is expensive to replace existing equipment unless there are distinct

advantages of using IPv6.

It is difficult to ensure 100 per cent compatibility with existing routers.