A Seminar Report on

LATEST TRENDS IN ASYNCHRONOUS TRANSFER MODE

Submitted to
JAWAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY,
ANANTAPUR
In partial fulfillment of the requirement for the award of the degree of
BACHELOR OF TECHNOLOGY
In
ELECTRONICS AND COMMUNICATION ENGINEERING

By
GR ANIL KUMAR (11691A0403)

DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING

MADANAPALLE INSTITUTE OF TECHNOLOGY AND SCIENCE
(Approved by AICTE, New Delhi, Affiliated to JNTU, Anantapur)
Madanapalle-517325, Andhra Pradesh.

MADANAPALLE INSTITUTE OF TECHNOLOGY & SCIENCE

(Approved by AICTE, New Delhi, Affiliated to JNTU, Ananthapur)
DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING

BONAFIDE CERTIFICATE
This is to certify that the technical seminar report entitled, LATEST
TRENDS IN ASYNCHRONOUS TRANSFER MODE, is a bonafide
work done by GR ANIL KUMAR (11691A0403) under our guidance
and supervision, in the partial fulfillment of the requirements for the
award of the degree of Bachelor of Technology in Electronics &
Communication Engineering in Madanapalle institute of Technology &
Science, Madanapalle, affiliated to Jawaharlal Nehru Technological
University, during the academic year 2014-2015.

Seminar Guide
Mr. D Balakrishna Reddy
Assistant Professor

Dept of ECE

Prof. A R REDDY, M.Tech, Ph.D.

Head of the department

Dept of ECE

Acknowledgement

I extend my sincere gratitude towards Mr. D Balakrishna Reddy, Asst Prof Dept. of Electronics
and Communication Engineering for giving me his valuable knowledge and wonderful technical
guidance. I also thank all the other faculty members of ECE department and my friends for their
help and support.

ABSTRACT

The paper begins with what is asynchronous transfer mode, ATM is simply a Data Link Layer
protocol. It is asynchronous in the sense that the recurrence of the cells containing from an
individual user is not necessarily periodic. It is the technology of choice for evolving B-ISDN
(Board Integrated Services Digital Network), for next generation LANs and WANs. ATM
supports transmission speeds of 155 Mbits/sec. In the future, Photonic approaches have made the
advent of ATM switches feasible, and an evolution towards an all packetized, unified, broadband
telecommunications and data communication world based on ATM is taking place.

CONTENTS
1. Introduction
2.

Basic Concepts
2.1 Connectionless versus Connection-oriented
2.2 B-ISDN
2.3 ATM

3.

Applications using Connectionless Communications

4. Trends
4.1 Networking is Critical
4.2

Peak of Technology Life Cycle

4.3 Standardization
5. Past Failures and Successes
6. Requirements for Success
7. Challenges
8. Summary

1 INTRODUCTION
ATM is simply a Data Link Layer protocol. It is asynchronous in the sense that the recurrence of
the cells containing from an individual user is not necessarily periodic. It is the technology of
choice for evolving B-ISDN (Board Integrated Services Digital Network),for next generation
LANs and WANs. ATM supports transmission speeds of 155 Mbits/sec. In the future, Photonic
approaches have made the advent of ATM switches feasible, and an evolution towards an all
packetized, unified, broadband telecommunications and data communication world based on
ATM is taking place.
These computers include the entire spectrum of PCs, through professional workstations upto
super-computers. As the performance of computers has increased, so too has the demand for
communication between all systems for exchanging data, or between central servers and the
associated host computer system. The replacement of copper with fiber and the advancements in
digital communication and encoding are at the heart of several developments that will change the
communication infrastructure. The former development has provided us with huge amount of
transmission bandwidth. While the latter has made the transmission of all information including
voice and video through a packet switched network possible.
With continuously work sharing over large distances, including international communication, the
systems must be interconnected via wide area networks with increasing demands for higher bit
rates. For the first time, a single communications technology meets LAN and WAN requirements
and handles a wide variety of current and emerging applications.

ATM is the first technology to provide a common format for bursts of high speed data and the
ebb and flow of the typical voice phone call. Seamless ATM networks provide desktop-todesktop multimedia networking over single technology, high bandwidth, low latency network,
removing the boundary between LAN WAN.

2 BASIC CONCEPTS
A lot of effort is currently being invested in the design of broadband communication systems.
The design of a system for supporting connectionless communications should be seen as an
integral part of these activities. In this chapter we review some basic concepts in B-ISDN,
discuss applications that will use a connectionless service, and survey related work. The purpose
of this chapter is to introduce the information that forms a starting point for the work presented
in the rest of the dissertation.
The concept of connectionless communication is essential to this work, and can only be
understood in relation to its opposite, i.e., connection-oriented communication. The network we
are considering for providing connectionless data communication is the B-ISDN. we will briey

review some basic concepts of ATM, which is the network technology that will be used to
support connectionless data communication.
2.1 Connectionless versus Connection-oriented:
In this dissertation the terms connectionless and connection-oriented are used as a characteristic
of a service, i.e., the observable behavior of a communication system. A service is connectionless
if two or more users of the service can transfer data using the communication system without
rst establishing and later releasing a connection.

Figure 2-1: Sequence of Service Primitives for a Connection-oriented Service

A service is connection-oriented if users must establish a connection before they can transfer data
using the communication system. The establishment of a connection is a negotiation between the
users who wish to communicate and the communication system. During the negotiation, state
information related to the connection is exchanged between the parties. The communication
system reserves resources for a connection, e.g., bandwidth.
The essential interactions between a communication system and its users can be described by
means of the sequence of service primitives, which are exchanged at the boundaries between the
system and its users. Figure 2-1 shows a likely sequence of service primitives in case of a
connection-oriented service, where two users want to communicate. Service primitives are
denoted by double arrows, to indicate an interaction in which both the user and the
communication system are involved. First, the initiating users request the system to establish a
connection (Connection Request). In this request at least the address of the required destination
and often parameters regarding the characteristics of the trafc and the required Quality of
Service (QoS) are passed to the system. The system analyses the parameters of the request and

determines to what extent it can support the requested connection. If the connection can be
supported, a Connection Indication primitive occurs between the system and the called user. This
user checks whether and under what conditions it wants to accept the connection. The acceptance
of the connection is indicated to the system by another interaction (ConnectionResponse).
Finally the initiating user is informed of the successful establishment of the connection
(ConnectionConrm). From now on, the users can transfer data. The data is passed from the
sending user to the system by means of a DataRequest primitive, transported by the system, and
passed from

Figure 2-2: Sequence of Service Primitives for a Connectionless Service

The system to the receiving user by means of a Data Indication primitive. After all data has been
transferred, the connection should be closed again. Therefore, one of the users, not necessarily
the initiating user, informs the system that it wants to release the connection (Release Request).
The system releases the connection and informs the other user (Release Indication).
For a connectionless service, the sequence of service primitives is much simpler (Figure 2-2). No
primitives for establishing or releasing a connection are needed. Each data unit is transferred
individually. The sending user passes the data to the system by means of a Data Request
primitive. The system transports the data and passes it to the receiving user by means of a Data
Indication primitive. All information regarding the transfer of the data unit, e.g., required
destination address, should be passed between the users and the network in these primitives.

An essential characteristic of a connectionless service is that the system does not have any
advance knowledge about the data that should be transferred (e.g., destination, or rate at which
data units are offered to the system). In case of a connection-oriented service this information is
agreed upon during the connection establishment, so that the system can reserve resources for the
transfer of the data units.
For both types of service there are classes of applications that they support best. Applications that
require a direct association between the users, such as telephony, are best served by a connectionoriented service. Other applications, such as those which involve the transfer of only a single unit
of information between a source and a destination are better served by a connectionless service.
Moreover, there are a huge number of applications that use the TCP/IP protocol suite. These
applications require a connectionless service at the network layer, also when they will be
supported by the B-ISDN . It is possible to use different types of services at different layers in
the network. In [82] the use of connectionless protocols at the network layer, to support all types
of applications is advocated. Others prefer to use a single connection-oriented protocol, i.e.,
ATM, to support all applications. Since ATM has been standardized for future integrated
networks, we assume ATM as the basic method to serve all applications. However, we believe
that additional provisions have to be made in the network to accommodate applications that are
connectionless in nature.
2.2 B-ISDN:
The Broadband Integrated Services Digital Network (B-ISDN) is expected to be the major future
telecommunications network for the wide area. It will provide a wider range of possible
applications, and support much higher throughputs than present telecommunication networks.
Applications, which can be very diverse in nature, will all be supported by a single network.
Moreover, a user can access the network via a single standardized interface, the User-Network
Interface (UNI).
The applications that should be supported by the B-ISDN are not only very diverse; they also put
very diverse requirements on the network. The network should support:
point-to-point as well as multi-point communications;
single-medium as well as multimedia communications;

 connectionless as well as a connection-oriented communications;

narrowband as well as broadband communications;
communications involving constant bit rate as well as variable bit rate trafc; and
communications for applications with a very diverse range of QoS requirements.
Some examples of applications that are foreseen for the B-ISDN are video-conferencing, High
Denition Television (HDTV) distribution, video-on-demand, telephony, and applications that
are currently supported by the Internet.
In order to support all these applications, the Asynchronous Transfer Mode (ATM) has been
adopted for multiplexing and switching in the network ([16]). ATM is a technique, where all user
information is transferred in Protocol Data Units (PDUs) of xed size, called cells. Cells are
identied by means of a header

Figure 2-3: The B-ISDN Protocol Reference Model

so that they can be identied for (de)multiplexing and switching. ATM should also be used at the
UNI of the B-ISDN. Two different UNIs have been dened ([78]); one with a bit rate of 155.520
Mbits/s, and one with a bit rate of 622.080 Mbits/s.
A Protocol Reference Model (PRM) has been dened for the B-ISDN in [18] (see Figure 2-3).
The PRM is divided into planes and layers. The division into (vertical) planes is done to identify
different types of functionality, i.e., for the transfer of user information, for the control of calls
and connections, and for management. The division into (horizontal) layers is done to create a
stepwise independency between the medium, used for the transmission of signals, and the
applications. A layer uses the next lower layer to provide a certain, less medium dependent, and
more application dependent, service to the next higher layer.
The B-ISDN PRM distinguishes between three planes. The user plane contains the functions that
deal with the user information. The control plane contains functions for the control of calls and
connections, and the transport of information on behalf of these functions (signaling). Finally, the
management plane provides for the coordination between user and control plane, and for the
management of individual entities and the overall network. The management plane has been
subdivided into layer management, performing management functions for specic protocol
layers and their entities, and plane management, performing coordination between the planes
and management of the system as a whole.
Within the user and control plane, protocol layers are identied. The lower layers are common to
both planes. The Physical Layer provides for the convergence of ATM cells to signals that can be
transferred over a physical medium. The ATM Layer provides for the end-to-end transfer of cells
along a connection. It performs switching and multiplexing of cells from different connections.

The ATM Adaptation Layer (AAL) adapts the common service provided by the ATM Layer to a
service that can better support specic classes of applications. It provides for instance for
segmentation and reassembly and for synchronization of source and destination.
Different views exist for relating the layering adopted in the B-ISDN PRM and the layering of
the Reference Model for Open Systems Interconnection (OSI-RM) ([65], [110], [111]). In the
context of this dissertation, it is convenient to situate the AAL in the OSI Data link Layer. The
ATM Layer can be considered as an upper sub layer of the OSI Physical Layer, or as a lower sub
layer of the Data link Layer.
The applications to be supported by the B-ISDN have very diverse communication requirements.
The ATM Layer is common to all applications. The purpose of the AAL is to adapt the ATM
service to the service required for the specic applications requirements. In order to come up
with a limited number of AAL services, applications have been classied according to the
following communication requirements:

timing relation between source and destination (required or not required);

bit rate (constant or variable); and
connection mode (connection-oriented or connectionless).

Since not all combination of the above requirements are foreseen, only four service classes are
distinguished according to Table 2-1. Clearly, the class of AAL service we are interested in for
the support of connectionless data communications over ATM is service class D. This class does
not provide a timing relation between source and destination, supports variable bit rate trafc,
and is connectionless.

Table 2-1: AAL Service Classes

For each of the service classes, one or more protocols have been dened that can provide the
service. For class D, two protocols, called AAL 3/4 and AAL 5 have been dened ([20], [75])1.
2.3 ATM :
The Asynchronous Transfer Mode (ATM) is a technique that is used for switching and
multiplexing in the B-ISDN. It can route xed-size data units, called cells, through a network of
switches interconnected by links, from source to destination. ATM is connection-oriented, i.e.,
prior to the transfer of cells, a connection is established through the network, and cells are
forwarded along the connection subsequently.

Multiplexing of cells from different connections on an ATM link is done using asynchronous
time division. Time is divided into slots, in which a single ATM cell ts. Slots can be assigned to
connections asynchronously, i.e., whenever a slot is needed for a connection it can be used.
Identication of the connection a cell belongs to is done by means of a label in the header of the
cell. Figure 2-4 gives an example of a series of cells, which are transmitted on a link
consecutively. All cells with the same label, e.g., 7, are identied as belonging to a certain
connection.

Figure 2-4: Asynchronous Time Division Multiplexing

The alternative to asynchronous time division multiplexing is synchronous time
division multiplexing, which is used in present telecommunication networks.
Time is again divided into slots, and slots are grouped into frames. Slots are
assigned to a connection synchronously, i.e., a connection is assigned the same
number of slots in each frame. Identification of the connection a slot belongs to is
done by means of the position of a slot in the frame. Figure 2-5 gives an example

of a series of consecutive cells in this case. All cells which are the third one of a
frame are identified as belonging to a certain connection.
An important advantage of asynchronous over synchronous time division multiplexing
is that the transmission capacity assigned to a connection can vary in
time, depending on the needs of the application. Furthermore, unlike with
synchronous time division multiplexing, where the assigned capacity is an
integer multiple of the smallest possible capacity (one slot assigned per frame),
any capacity can be assigned to a connection.
ATM switches receive cells on a number of incoming links. The connection an
incoming cell belongs to is uniquely defined by the incoming link, and the label
carried in the cell header. In the switch a routing table is maintained, in which the
outgoing ATM link, and the required label for the connection on that ATM link is
given for each connection. Using this table, the switch can replace the label of a
cell, and forward it to the proper outgoing link. Figure 2-6 gives an example of
ATM switching. The shaded entry in the routing table indicates that all cells that
arrive on link 2 with label 7 should be forwarded to link 4, while the label should
be modified to 3. ATM cells are drawn as in Figure 2-4, i.e., from left to right, first
the header, which contains the label, and then the payload.
As stated before, ATM is a connection-oriented technique. The header of an ATM
cell relates the cell to a previously established ATM connection. Two types of
ATM connections are identified, Virtual Path Connections (VPCs) and Virtual
Channel Connections (VCCs). Therefore, also two identifiers can be found in the
header of the cell, a Virtual Path Identifier (VPI) and a Virtual Channel Identifier
(VCI). The combination of the two, referred to as VPI/VCI, determines the ATM
connection a cell belongs to.
A physical link between two switches carries a number of Virtual Path Links
(VPLs), each of them is identified by a VPI. The concatenation of a number of
VPLs forms a Virtual Path Connection (VPC). Within a VPC, a number of Virtual
Channel Links (VCLs) can be identified. Each VCL is identified by a VCI, which is

Figure 2-6: ATM Switching

li

unique for that VPC. The concatenation of a number of VCLs forms a Virtual
Channel Connection.
Figure 2-7 shows the relationship between the different types of links and connections.
In the figure, a series of 7 ATM switches is shown. On a link between a pair
of switches, a number of VP links can be identified. A concatenation of VP links
forms a VP connection, e.g., from the first to the third switch. All switches operate
on VPs; only some of them operate also on VCs. These are also visible at the VC
level. Within a VP connection (between two VC switches), a number of VC links
can be identified. The concatenation of a number of these links forms a VC
connection, e.g., between the first and the seventh switch.

An ATM cell has a payload field of 48 octets. The header of the cell is 5 octets
long. The major fields in the header are the VPI and VCI field. Furthermore, the
header contains a Header Error Control (HEC) field to protect the header against
bit errors. A Cell Loss Priority (CLP) field can be used to indicate the relative
importance of the cell in the connection. The Payload Type (PT) field carries additional
information, e.g., an ATM-user-to-ATM-user indication that is transported
transparently by the network. At the UNI, also a Generic Flow Control (GFC)
field is present in the header in order to control the access to a shared link.

Figure 2-7: ATM Connections

Error control should be performed in layers above ATM on an end-to-end basis.

Only errors in the header are detected or corrected (depending on the mode of
operation) by means of the 8 bit HEC field, if possible.
In order to protect an ATM network against overload and guarantee the
requested QoS to the users, bandwidth is reserved for each connection. A
Connection Admission Control (CAC) mechanism checks whether or not bandwidth
for a newly requested connection is available. If not, the connection is
refused. If the bandwidth is available, the connection is established, and the
bandwidth is reserved. In order to check if the user does not violate the agreed
bandwidth another mechanism, called Usage Parameter Control (UPC), has to be
implemented at the border of (the public part of) the network, directly after the
UNI. Cells that cause a violation of the agreed bandwidth are either discarded
directly, or given a low cell loss priority, so that they are the first ones to be
discarded in case of congestion.

3 Applications using Connectionless Communications

The emphasis in this dissertation is on connectionless communications over the
wide area, with throughputs that have not been available up to now. In order to
provide insight into the requirements on the system, we will explore the characteristics
of applications that have been identified in a number of publications, e.g.,
[5], [39], [95], [96], [106], [128], [146]:
file transfer,
terminal access,
information retrieval (e.g., World Wide Web),
computer graphics,
distributed supercomputing,
remote procedure call (RPC),
virtual memory page swapping and paging, and
electronic mail.
Although not all applications listed here need necessarily be supported by a

connectionless network, most of them are connectionless in nature. This means

that the applications need to transfer one or more individual pieces of information,
not a stream of related pieces of information. Electronic mail is an example
of an application that is inherently connectionless. Its purpose is to transfer individual
messages from source to destination. There is no need for an association
between source and destination before the message is transferred.
The end-systems that implement the applications can be attached directly to the
B-ISDN. However, most applications have been first introduced in a local environment,
so that a lot of existing end-systems have been attached to Local Area
Networks (LANs). Another reason for attaching end-systems to LANs is the
locality of a large share of the generated traffic.
However, there is a growing demand for communications over a wider area. This
can be accommodated by interconnecting LANs by means of networks that span
a wider geographical area. Most LANs use connectionless protocols. In order to
avoid complicated and costly protocol conversion, it is often desirable to interconnect
these LANs by means of connectionless networks. A first stage into this

Figure 3-1 Connection of End-systems, LANs, and MANs to a WAN

development is the interconnection of LANs by Metropolitan Area Networks

(MANs), which typically have a geographic span of about 50 kilometers. Many of
these MANs are based on the Distributed Queue Dual Bus (DQDB) technique,
defined in the IEEE 802.6 standard ([63]).
If a Wide Area Network (WAN) supporting broadband connectionless communications
becomes available, it can be used to interconnect these MANs, and to
interconnect LANs directly. Furthermore, end-systems with high communication
needs can be attached directly to the WAN. This will lead to a scenario where
end-systems are attached to either a LAN, a MAN, or a WAN, with LANs being
connected to either a MAN or a WAN, and the MANs connected to the WAN (see
Figure 3-1).
We refer to the systems that interconnect different networks, and perform the
needed conversion, as Interworking Units (IWUs). These IWUs can function as
routers, interconnecting the different networks at the network layer, e.g., using
the Internet Protocol (IP) ([107]) or the ISO Connectionless Network Protocol

(CLNP) ([66]). The IWU can also function as a bridge, interconnecting the
different networks at the medium access control (MAC) sub layer ([128]).
It is expected that most traffic to be transported by the wide area connectionless
network will initially come from end-systems attached to LANs. Interconnection
of these LAN is often referred to as the most important application of the
network. In fact, the real applications to be supported are those listed above.
From the point of view of the WAN, the LAN only aggregates the traffic generated
by these applications.
In order to get more insight in the traffic generated by the individual applications,
we consider the requirements on the transfer of an single unit of information. [39]
and [95] list for a number of applications the expected size of an information unit.
Furthermore, they give an indication of the required response time for these
applications. Figure 3-2graphically represents these requirements. For each
application an area of requirements is given, expressing both the uncertainty and
the variation in requirements. Note that the information units sizes and response
time requirements are given at the application level. An information unit may
very well be transported in a number of smaller packets at the network level

Figure 3-2 Applications Requirements

Figure3-3 Throughput Requirements of Applications

In Figure 3-2 a number of diagonal lines have been drawn that give an indication
of the throughput required to transfer information units of a certain size within
the response time. Figure 3-3summarizes these throughput requirements. It can
be seen from the figure that throughput requirements range from less than a
kilobit per second for electronic mail and terminal access to a gigabit per second
for computer graphics.
In principle, the given applications do not tolerate any loss or corruption of data.
This does not imply that no loss or corruption can be tolerated from the connectionless
service we are designing. End-to-end protocols used on top of the
connectionless service (e.g., a transport protocol) can enhance the provided QoS.

4 .1 Networking is Critical :

Networking has become the most critical part of computing. Today, computers are used
mostly for transferring information from one peripheral to another, from network to the
disk, from disk to the video screen, from keyboard to the disk, and so on. Mail, _le transfer,
information browsing using World Wide Web, Gopher, and WAIS takes up more time of the
computing resources than computing per se. Initially, when the computers were designed,
the performance was measured by the \add" instruction time. Today, it is the \move"
instruction that is the key to the perceived performance of a system. This means that the
bus performance is more important than the arithmetic logical unit (ALU) performance.
I/O performance is more important than the SPEC marks.
There are several other reasons for communications and networking becoming critical. First,
the users have been moving away from the computer. In the sixties, computer users went
to computer rooms to use them. In the seventies, they moved to terminal rooms away from
the computer rooms. In the eighties, the users moved to their desktop. In the nineties,
they are mobile and can be anywhere. This distance between the users and the computers
has lead to a natural need for communication between the user and the computer or the user
interface device (which may be a portable computer) and the servers.

Second, the system extent has been growing continuously. Up until eighties, the computers
consisted of one node spread within 10 meters. In nineties, the systems consists of hundreds
of nodes within a campus. The increasing extent leads to increasing needs for computing.

Figure 4-1 Shaped technology curve

from timesharing to personal computing. Now we need ways to work together with
other users. So, in the next ten years, emphasis will be on cooperative computing. This will
further lead to increase in communications.
In the last decade, we were busy developing corporate networks, and campus networks. In
the next decade, we need to develop intercorporate networks, national information infrastructures, and international information infrastructures. All these developments will lead to
more growth in the field of networking and more demands for the personnel with skills in
networking.
The increasing role of communications in computing has lead to the merger of the telecommunications and computing industries. The line between voice and data communications is
fading away. Data communication is expected to take over voice communication in terms of
volume.

4.2 Peak of Technology Life Cycle

Most technologies follow an S-shaped curve shown in Figure 1, where the number of problems
solved is plotted against time. There are three distinct stages in the life of a technology.
In the beginning, all problems are hard and it takes a lot of resources and time to solve
a few problems. At this stage, a lot of money is spent in research but there is very little
revenue. Most of this research is funded by the traditional government funding agencies,
such as, National Science Foundation and Advanced Research Project Agency (ARPA) in
the United States.
After some of the key problems have been solved, a lot of other problems can be solved by
spending little money. At this stage, the curve takes an upturn. The amount of revenues to

Figure 4-2 The exponential growth of Internet

be made from the technology is much more than the investments. It is at this stage, that
the industries take over technology development. Numerous small companies are formed and
quickly grow to become large corporations.
Finally, when all the easy problems have been solved, the remaining problems are hard and
would require a lot of resources. At this stage, the researchers usually move on to some other
technology and a new S-shaped curve is born.
The computing industry in general and the networking sector in particular is currently going
through the middle fast growing region of the technology life cycle curve. The number of

problems solved is indicated by the deployment of the technology. In case of networks, one
can plot the number of hosts on the networks, bytes per host, number of networks on the
internet, total capacity (in MIPS) of hosts on the network, total memory or total disk space
of the hosts on the network. In each case, one would see a sudden exponential up turn in
the last few years.
Figure 2 shows the famous Internet growth curve. The figure shows the number of hosts
on the Internet in the last 20 years. The data before July 1988, although plotted is hardly
visible. Since 1988, growth has defied all predictions.

4.3 Standardization
When a technology reaches the middle fast growing region, it becomes necessary to standardize it to make it usable for the masses. The computing industry in general and networking
in particular is undergoing through this phase. Even if people use different computers, it
is necessary that the networking interfaces be standardized so that these diverse computers
can communicate with one another.
The standardization requires a change in the way business is done. Before standardization,
a majority of the market is vertical. The only way for users to maintain compatibility is to
buy the complete system from one manufacturer. System vendors make more money than
component vendors. IBM, DEC, and Sun Microsystems are examples of such system vendors.
After standardization, the business situation changes. Users can and do buy components
from different vendors. The market becomes horizontal. Companies specializing in specific
components and fields take prominence. Intel for processors, Microsoft for operating systems,
Novell for networking are examples of this trend.
To survive in this post-standardization era, invention alone is not sufficient. Only those new
ideas that are backed by a number of vendors become standardized and are adopted. It,
thus, becomes necessary to form technology partnerships.

5 Past Failures and Successes

In the last fifteen years, we have seen a number of networking technologies that were very
promising during their life time but were not successful in the long run. A sample of such
technologies is listed in Table 5-1. In each case, the technology listed in the second column was
more promising than the one in the third column until the year shown in the first column.
In early 80's, when Ethernet was being introduced, some argued that broadband Ethernet,
which allows voice, video, and data to share a single cable would be more popular than
baseband Ethernet. As we all know, today there are a few broadband installations. Most
installations of Ethernet are baseband. The cost of combining the three services was just
too high. The analog circuits required for frequency multiplexing were not as reliable and
economical as digital circuits with separate wiring.
Around the same time, when computer companies were trying to sell Ethernets, PBX manufacturers were presenting PBX as the better alternative, again because it was already there
and it could handle voice as well as data. However, PBX was not accepted by the customers
simply because it did not provide enough bandwidth.
The Integrated Service Digital Network (ISDN) was standardized in 1984 and was very
promising then. However it's deployment has been much too slow. Even after ten years, it
is not possible to get an ISDN connection at most places. Even at those places where it is
available, the 64 kbps bandwidth provided by it is not sufficient for most data applications.
For low bandwidth applications, modems on analog lines provide a better alternative. Modem technology has advanced much beyond expectation. Today, one can get 28.8 kbps and
56 kbps modems that work with all pervasive analog lines and do not have monthly charges
associated with the extra ISDN line.
In 1986, IEEE 802.4 (token bus) was touted as a better alternative than IEEE 802.3/Ethernet

Table 5-1 Networking Failures and Successes of the Past

for real time environment. It was said that Ethernet could not provide the delay
guaranteesrequired for manufacturing and industrial environments. Manufacturing Automation
Proto-col (MAP/TOP) was seen as the right solution. Today, IEEE 802.3/Ethernet is used in
allsuch environments. Token buses are practically nonexistant.Up until 1988, ISO/OSI protocol
stack was seen as the leading contender for networking
everywhere. Networking researchers in most countries were implementing ISO/OSI protocols. The United States Government Open Systems Interconnection Prfile (GOSIP) even
made ISO/OSI a mandatory requirement for government purchases. Today, TCP/IP protocol stack dominates instead. The OSI protocols suffered from the common problems of
standards: it had too many features. Any feature required by any application in the world
needed to be supported by the standard. The protocols took too long to standardize and
were quite complex. The \build before you standardize" philosophy of the TCP/IP protocol
stack helped in its success.
Up until 1991, IEEE 802.6 Dual Queue Dual Bus (DQDB) was seen as a promising candidate
for metropolitan area networks. It is no longer considered viable. The unfairness problem
and general problems of bus architectures have made it undesirable.
Xpress Transfer Protocol (XTP) was designed as the high-performance alternative to TCP/IP.
Protocol Engines { the company leading the design of XTP declared bankruptcy in 1992.

6 Requirements for Success

There are several lessons to learn from the list presented in Table 1. First, all technologies
appear very promising when first proposed. However, not all survive. Those that survive
meet all the following requirements.
1. Low cost
2. High performance
3. Killer application
4. Timely completion
5. Interoperability among various implementations of the same technology
6. Coexistence with existing (legacy) technology.
After a brief overview of ATM networks in the next section, we discuss some of these issues
as they apply to ATM networks.

Challenges

In this section, we discuss a few of the requirements identified earlier and see
what needs to be done to ensure success of the ATM technology.
7.1 Economy of Scale
Today, networking technology seems to be far ahead of the applications. High-speed fibers
have been installed but there is not enough video traffic to fill them. Invention is becoming
the mother of necessity. We need to create a need for the high bandwidth. In such a
situation, generally low cost is the primary motivator. When the high-speed technology
is proportionately cheaper than lower-speed alternatives, the buyer's considerations change
from \Why should I buy high speed?" to \Why should I not buy high speed?"
Today, there is diseconomy of scale. Higher speed networks cost more in per-bit than lower
speed networks. Ten Mbps Ethernet cards can be had for $50. However, 100 Mbps FDDI

cards cost closer to $1000. This diseconomy of scale has a significant impact on user adoption.
We have seen this happen in other areas of computing. Today ten 100-MIPS computers cost
much less than one 1000-MIPS computer. Therefore, we see more distributed computing
than supercomputing. Applying the same logic, it appears that unless there is economy of
scale, people may divide their networking applications among multiple low-speed links than
one high-speed link. Of course, there are a few applications that will not work on speeds in
the range of 10 Mbps. For these, the users have no choice but to use higher speed links.
This diseconomy of scale a_ects all high speed technologies including ATM. However, ATM
has a bigger uphill battle due to its newness. In a recent ATM Forum user survey conducted
Dr. John McQuillan, the users were asked that given the same price, which 100 Mbps network
they would buy: ATM or 100-Mbps Ethernet. The answer was Ethernet because it is
something the users feel very comfortable with. We ATM designers will have to work hard
to get the ATM equipment prices below the 100-Mbps Ethernet to get acceptance.

7.2 Tariff
Tariffing the ATM tra_c is another problem. Today's telecommunications tari_s are designed for low bandwidth high cost voice traffic. It costs $25/month for a simple 64-kbps
analog phone line. At this rate, the phone company has the potential of making $211k/year
on a 45 Mbps link. A coast to coast T3 link does cost $180k to $240k per year. A 155 Mbps
link would cost three times as much. However, 155 Mbps ATM circuits are being tari_ed at
$13k to $45k per year. That's 10 to 50 times cheaper than today's rates. The situation is
similar to that of the computer industry. When the computer prices started going down, the
old established companies designed to sell expensive mini and supercomputers had trouble
keeping their overheads down to be able to sell personal computers. While new companies designed to make money selling personal computers cheaply ourished, old companies
tried hard to maintain their existing business almost to the point of bankruptcy. Success of
cheap telecommunications using ATM technology has the potential of doing the same to the
telecommunications industry. The telecommunications companies with overheads designed
for expensive voice services will have a tough time selling cheap data services. The danger

of decreasing revenues may prove to be a hinderence to the success of ATM at least in the
wide area networking (WAN) market.
7.3 Performance
Figure 3 shows a layered model of people involved in designing a high speed network. At
the bottom are the physicists, who work on _bers and lasers. They are working today with
10 Gbps lasers and their challenge is to design 100 Gbps lasers. The next level up are
the LAN designers who are designing FDDI and ATM networks using these _bers. The
LAN designers are working at 100 to 155 Mbps { two orders of magnitude lower than the
physicists. Then there are LAN adapter designers who take the 100 Mbps protocol standard
and design adapters which run barely at 20 Mbps. Although some users aren't aware of this,
many FDDI adapters cannot transmit or receive more than 20 Mbps. Thus, we loose the
performance by a factor of 5 at this layer.

Figure 7-1 A layered view of people who affect the performance seen by the user
The next (fourth) layer consists of processor designers. Unless designed carefully, a processor

may not be able to keep up with the high speed LAN adapters. The fifth layer of operating
system reduces the performance further. The sixth layer is that of network protocols{ some
of which are not able to cope with high speed links. Finally the top (seventh) layer is the
application, which sees a usable bandwidth of a few megabits per second.
Unless higher layers are improved, changing the lower layers will not result in any higher
performance for the user. Today's problem is not so much in networking protocols as it is in
proper I/O designs for the processors and operating systems. We can provide the user with
155 Mbps or 622 Mbps links, but they will not be able to use it unless, operating system
and processor designs are improved accordingly. The only exception is the backbone where
specialized hardware and software are used. The backbone components (switches, routers, or
bridges) are designed specially for high communication speeds. Thus, the _rst place where
high-speed is will be used is in the backbone. The desktop market will have to wait for
better operating systems and processors.
Another lesson to learn from this layered model is that for high speed networks to become
a reality, all seven layers have to be improved. Bad performance even in one layer can delay the
introduction of the High-speed Networking.
7.4 Application
It is well known that future applications will be multimedia including data, voice, image, and
video. What is less well known is that the voice traffic will be a negligible part of the traffic.
This can be seen by considering what happens on telephone networks today. On the AT&T
network, approximately 125 to 130 million calls are made per day and an average call lasts
around 5 minutes. Each call requires a bandwidth of 64 kbps. Thus, the total bandwidth
used by voice in AT&T network is approximately 28.8 Gbps, which is only one thousandth
of the potential bandwidth of a fiber. Even if all 200 million people of the United States were
to talk 24 hours a day, the total bandwidth required will be only 12.8 Tbps, which again
is less than the potential bandwidth of a fiber. In a survey of private networks reported in
the August 1992 issue of IEEE Spectrum, it was found that in 1985, 75% of the traffic was
voice. In 1990, the voice percentage dropped to 56% and in 1995, it is projected to be 39%.
If we were to make a forecast based on this trend, we would conclude that the voice traffic

will be zero or negligible by the year 2010. The reason for this decline is not that people are
not talking enough but that while voice traffic is limited by the population, the data traffic
is not. Computers have no limit on the speed at which they can transmit and so there is
no limit to the value to which data traffic can grow. It is this exponentially growing data
business that most telecommunications companies want to get into.
Next, let us consider characteristics of video traffic. One hour of uncompressed HDTV
requires 540 GB of storage. At today's storage prices of $1/MB, this works out to approximately $150 per second of video. This is somewhat expensive and only researchers funded
on government grant can a_ord to store the video at this price. If compressed, the storage
requirements drop by a factor of 60 to 200 and the price becomes $2.50 to $0.75 per second,
which is more reasonable. The conclusion is that most video will be in compressed form simply for storage. Compression means that the bandwidth requirements vary, and therefore,

variable bit rate (VBR) service rather than constant bit rate (CBR) is likely to be used more
often.
Also, at high speeds, the connection holding times become shorter. At 1 Gbps, it takes
only 10 seconds to transmit 1 hour of compressed VHS movie. It takes even smaller time at
higher speeds. Thus, unless the bandwidth is free, most users of high speed will start and
shutdown the connection after 10-20 seconds. In other words, the traffic will be short-lived
and bursty. This is closer to today's data traffic than voice traffic. For ATM networks to
succeed, they should be able to handle the bursty traffic eficiently.
In 1984, when the ATM cell size was being decided in CCITT, they were thinking about 64
kbps voice. At that speed, 32-byte cells need 4 ms. If larger cells were used, the time to
collect the voice would become too large and would require echo cancellation. The Europeans,
therefore, wanted 32-byte cells, while US position was that the cells be at least 64 bytes long.
The limit of 48 bytes was chosen as an average of 32 and 64. In other words, the cell size
was chosen not for high speed applications but for 64-kbps voice applications. Several other
design and implementation decisions for ATM networks were similarly done as if they were
being designed for voice. One example of such design philosophy is simply dropping cells
on congestion. Requiring users to indicate which cells aren't important by the congestion-

loss priority (CLP) bit is another example. For voice, some cells can be dropped without
significant impact. However, this is not true for data. Every single bit is important and all
dropped bits have to be retransmitted.
The cell size is not suitable for high-speed applications in general and for video traffic in
particular. A single HDTV frame requires 50,000 cells. Switching 50,000 times for each
frame is a not the optimal way.
Prior to the formation of the ATM Forum in October 1991, most of the ATM networks
design decisions were made as if the network was being designed for voice. It is only in 1994
that the importance of data traffic was realized and the available bit rate (ABR) service was
introduced. It is now well understood that the key to ATM technology success is its support
of data traffic. If ATM fails to support data, it will not be able to stay around for video
traffic.

7.5 Scalability
Queueing theory tells us that the variance of response time in a queue depends upon the
variability of the service time (cell time) and square of the cell time:
Variance(response time) = Variance(cell time) + Cell time2
Making the cells same size makes the first term zero. Making the cell time small reduces the
variance in response time. For delay sensitive real time applications, smaller cells provide
reduced delay variation.
It is important to note that we used time in the above equation and not size. The time
requirements of an application do not change as the bandwidth of the network changes.
For example, 30 frames per second video will need one frame every 33 ms regardless of
the speed of the link. Even at Gigabit per second or Terabit per second speeds, the video
will need a response time variation in milliseconds. The cell time of 6 ms would satisfy
most delay sensitive applications. Although 6 ms is 48 bytes at 64 kbps, it is 900 kB at 1.2
Gbps. By using smaller cells at higher speeds we get micro- and nano-second delay variation.
Unfortunately our eyes cannot even feel the difference between millisecond and microsecond

delays, and therefore, we are wasting switching resources.

The cost of a switch depends partly upon its switching speed in cells per second. Given a
switch design, it is possible to make a higher speed switch by simply increasing the cell size
without any significant increase in cost.
What we need are \constant-time" cells and not \constant-size" cells for scalability to high
speed. With constant size cells, the cell time decreases as the speed increases and it becomes
necessary to switch more and more cells per second thereby increasing the cost.
The telecommunications industry claims SONET to be scalable in bandwidth. SONET uses
constant-time frames. All SONET frames are 125 microseconds long. As the speed increases,
the number of bytes in the frame increases proportionately.
Many of today's ATM networks use SONET links. In these networks, the large video image is
broken down in small cells, which are then packed into a large SONET frame and transmitted.
At the receiver, the SONET frame is unloaded, the information is switched cell by cell { all
of whom are probably going to the same destination. After switching, the cells are loaded
into another SONET frame and forwarded to the next switch. This process of unloading
SONET frames and switching cell by cell is clearly unnecessary, given that at high speed,
the amount of information to be transmitted is also generally higher. We could just switch
SONET frames or use a technology which makes use of the best features of SONET and
ATM.

7.6 Simplicity
The final challenge that ATM technology faces is that of keeping it simple. During the
design of IEEE 802.3/Ethernet standard, there was fierce competition between CSMA/CD
and token ring camp - both trying their best to keep their design more cost effective than
the other. This competition was good and did help keep the scope of both standards limited.
For ATM, unfortunately, there is no equal competition at this time. Thus, any thing that
needs to be done by networks has to be done by ATM networks. The design is becoming
too complex. Too many options are being added. ATM networks have to work for constant
bit rate (CBR) traffic as well as variable (VBR) and available bit rate (ABR) traffic. They

have to work for local area (LAN) as well as for wide area (WAN). They have to work for
low-speed as well as for high speed. They have to work for private networks as well as for
public networks. All these options add to the complexity. The situation is similar to that of
OSI.
One of the advantages of switches over routers was that switches were supposed to simple.
They no longer are. For ATM switches, switching is only a negligible part of their responsibility. A large part of switch resources is consumed by connection set up, route determination,
address translation, multicasting, anycasting, ow control, congestion control, and so on.
Another element adding to the complexity of ATM is the fact that it is being developed at
multiple standards bodies: ITU and ATM Forum. Strictly speaking, ATM forum is not a
standards body. ITU is supposed to develop the standards and ATM Forum is supposed to
select a subset of options provided by ITU. But in reality, ITU is too slow. ATM Forum
cannot wait for ITU to finalize its standard and so it is taking a leading role in developing
it in parallel. A considerable amount of time at both bodies is used in reconciling the
agreements made at the other body. Vendors will end up implementing both ITU and ATM
Forum versions of the standards and users will have to bear the cost even though just one
set would have been fine.
8 Summary
Networking is a critical part of computing today and is growing exponentially. Networking
is in the mid fast-growing region of the technology curve.
High speed networking will succeed if and only if there is economy of scale so that using
higher speed links results in cost savings. Unfortunately, this is not case right now. We face
the danger of users dividing their applications and using several low-speed links.
Considerable amount of resources are being put on ATM networks. However, its success
will depend upon our being able to transfer data at a lower cost and higher performance
than legacy LANs. Also, we will have to control the desire to incorporate all options at once
otherwise it will become too complex.

BIBLIOGRAPHY
[1] M. Ajmone Marsan, G. Conte, G. Balbo, A Class of Generalized Stochastic Petri Nets for
the Performance Evaluation of Multiprocessor Systems, ACM Transactions on Computer
Systems, Vol. 2, No. 2, May 1984, pp. 93 - 122.
[2] G.J. Armitage, K.M. Adams, Packet Reassembly During Cell Loss, IEEE Network, Vol. 7,
No. 5, September 1993, pp. 26 - 34.
[3] R. Atkinson, Default IP MTU for use over ATM AAL5, RFC 1626, May 1994.
[4] ATM Forum, ATM User-Network Interface Specication- Version 3.0, ISBN 0-13-2258633, PTR Prentice Hall, Englewood Cliffs, NJ, USA, September 1993.
[5] J.J. Barrett, E.F. Wunderlich, LAN Interconnect Using X.25 Network Services, IEEE
Network, Vol. 5, No. 5, September 1991, pp. 12 - 16.
[6] Bellcore TA-TSY-000772, Generic System Requirements in Support of Switched Multimegabit Data Service, Bellcore, Issue 3, October 1989, Supplement 1, December 1990.
[7] C. Blondia, O. Casals, Statistical Multiplexing of VBR Sources: A MatrixAnalytic
Approach, Proceedings of the workshop on Performance Aspects of ATM, PTT Research,
Leidschendam, The Netherlands, October 1991.
[8] G. Boiocchi, P. Crocetti, L. Fratta, M. Gerla, M.A. Marsiglia, ATM Connectionless Server:
Performance Evaluation, Proceedings IFIP WG6.4 International Workshop on Performance of
Communication Systems, Martinique, French Caribbean Island, January 1993, pp. 185 - 195.
[9] G. Boiocchi, P. Crocett, L. Fratta, M. Gerla, Performance Evaluation of a Connectionless
Multicast Service, Elsevier, Teletrafc Science and Engineering, Vol. 1, Proceedings of the 14th
International Teletrafc Congress - ITC 14,