Internet Traffic Engineering

362 Pages·2005·6.06 MB·English

by  Gunhild Luke

Contents
Internet Traffic Engineering
1 Guest Editorial; Terje Jensen
Telektronikk Section I: Introduction/Overview
Volume 97 No. 2/3 – 2001 3 Computers and Communication; Yngvar Lundh
ISSN 0085-7130
20 Internet Protocol and Transport Protocols; Terje Jensen
Editor:
Ola Espvik 39 Traffic Engineering Principles, Activities and Mechanisms; Terje Jensen
Tel: [phone]
email: [email protected] 54 Basic IP-related Mechanisms;Terje Jensen
Status section editor:
Section II: Traffic, Resources, Routing
Per Hjalmar Lehne
Tel: [phone] Planning and Designing IP-based Networks;
email: [email protected] Terje Jensen, Mette Røhne, Inge Svinnset, Rima Venturin and Irena Grgic
Editorial assistant:
107 Achieving Service Differentiation in a Differentiated Services Network by
Gunhild Luke
Use of MPLS; Inge Svinnset
Tel: [phone]
email: [email protected]
116 The Design of Optimal Multi-Service MPLS Networks;
Editorial office: Åke Arvidsson and Anthony Krzesinski
Telenor Communication AS
Telenor R&D 130 State-of-the-art of IP Routing;Boning Feng, Anne-Grethe Kåråsen,
PO Box 83 Per Thomas Huth and Bjørn Slagsvold
N-2027 Kjeller
145 Routing Strategies for IP Networks;
Norway
Tel: [phone] Walid Ben-Ameur, Nicolas Michel, Bernard Liau and Eric Gourdin
email: [email protected]
159 IP Multiplexing for Low Capacity Links?;Olav Østerbø
Editorial board:
Ole P. Håkonsen, Section III:Interdomain, SLA, Policy, Management
Senior Executive Vice President.
Oddvar Hesjedal, 170 Traffic Engineering – Inter-domain and Policy Issues; Terje Jensen
Vice President, R&D.
Bjørn Løken, 186 Agreements in IP-based Networks; Irena Grgic and Mette Røhne
Director.
213 Modelling the Topology of IP Networks;
Graphic design: Terje Henriksen, Anne-Grethe Kåråsen and Ståle Wolland
Design Consult AS, Oslo
Section IV: Measurements
Layout and illustrations:
Gunhild Luke, Britt Kjus (Telenor R&D)
230 Traffic Measurements in IP Networks;Brynjar Å Viken and Peder J Emstad
Prepress and printing:
245 A Distributed Test Environment for IP Performance Evaluation;
Optimal as, Oslo
Poul E Heegaard and Brynjar Å Viken
Circulation:
4,000 269 Methods for monitoring, controlling and charging QoS in IP Networks;
Jorma Jormakka, Irena Grgic and Vasilios Siris
Section V: Systems and Services
287 Network Principles and Applications;Terje Jensen
311 Voice Transmission over Internet;Johan M Karlsson
319 Quality Issues for Packet-based Voice Transport; Danny De Vleeschauwer,
Annelies van Moffaert, Maarten J C Büchli, Jan Janssen and Guido H Petit
332 Quality of Service in UMTS;Thor Gunnar Eskedal and Frédéric Paint
346 Optical Network Functionality: From “Dumb Fat Pipes”
to Bright Networking;Evi Zouganeli
355 Abbreviations
Guest Editorial
TERJE JENSEN
To survive in today’s competitive environment, The Internet Protocol (IP) has become a pivotal
a service provider must continually evolve its component in communication between various
network and enable new revenue-generating ser- devices. It is rarely possible to make a single
vices faster and more cost effectively than the protocol suffice the diverse needs of all applica-
competitors. Prior to the Internet, prior to recent tions and users. To a certain extent, however,
mobile services, prior to e-business, a change in one may claim that the IP suite is addressing
the telecommunication industry was seen as such an objective. However, looking at the origi-
more predictable. Business leaders knew to take nal use of IP when it was designed, there are
action – actions like reducing costs, launching many other applications of the protocol these
new products, upgrading the networks, and so days; as more demanding services – like tele-
forth. Now providers are far less sure who their phony, video distribution and mission-critical
Terje Jensen competitors are, the value of their core strengths business applications – are gradually put onto
and skills, and whether the business they have IP-based networks, additional functions must be
done well in for many years will continue to implemented in the networks and end-systems.
keep them profitable in the future. Hence, one is stretching the capabilities of IP
and additional mechanisms are necessary to
Some may claim that a main cause for the uncer- allow IP to hold on to a central position. Several
tainty is that recent development of applications of these mechanisms are related to Traffic Engi-
and service demands has been going on outside neering, that is, means activated to ensure the
the sphere of the service providers. In particular, performance of the communication solutions.
the Internet Engineering Task Force (IETF) has This will also allow for more predictable
received major contributions from some main responses on service requests and swiftly
players and been guided by inputs from the support of more advance services and users.
academic world. For sure this has resulted in a
plethora of applications and usage patterns, and Quite a few phenomena influence the evolution
phenomenal traffic growth. However, as more of IP-based networks, of which some interacting
Front cover: The conflict
commercial concerns are entering the stage the factors are:
between market require-
providers would regain more control, for exam- • Increased load and expansion; more efficient
ments and the service
ple by utilising the Traffic Engineeringsolu- ways of handling the traffic is sought. Scala-
provider’s resources
tions. This also advocates further work on stan- bility challenges are commonly faced for this
At low traffic pressure incom-
dardisation, ensuring interoperable configura- reason.
ing traffic equals traffic car-
ried. The artist Odd Andersen tions. Including procedures for managing multi-
indicates this situation by his ple service types and requirements, Internet • New technologies; efficient ways of interact-
45 degrees black lines. How- Protocol (IP) Traffic Engineering thus provides ing with IP-based networks are looked for.
ever, every traffic machine – mechanisms for optimal operation and manage- Anexample of this is the relation between
be it switches, routers or the
ment of the IP-based network. Thereby, a functions related to IP and an underlying
whole network – has a capac-
ity limit at which the line provider would also improve its chances in optical layer.
breaks into a horizontal posi- thefrenzied market.
tion. After that no more traffic • New user groups; additional requirements
can be carried whatever the Basically, one option could be simply to increase could be placed on the IP-based network.
pressure!
the capacity of the network, like adding more Thus, efficient ways of differentiation be-
But the enormous increase bandwidth to the links. A problem with this tween the groups are asked for, also accom-
–and non-established nature
argument is that capacity should then be added panied by appropriate charging solutions.
– of Internet traffic blasts
wherever there is a problem, including the pro-
through all aspects of capacity
and quality constraints. The cessing capacity, and also in the access network, • New applications; innovative ways of utilising
artist’s skyrocketing white on the servers, etc. Furthermore, and perhaps an IP-based networks are steadily observed, e.g.
lines indicate the unprece- even heavier argument is that the possibility for related to electronic business, mobile services,
dented demands to proper
service differentiation would then still be rather and so forth.
engineering of traffic ma-
limited. Being able to offer a portfolio of differ-
chines in thenew market
environments. ent services is recognised as a key enabler for • Increased dependency on the network; coming
ensuring a provider’s profitability. Again, from the service providers themselves as com-
The artist’s generic message:
Matching capacity of all com- Traffic Engineering is promoting a set of mecha- mercial aspects, but also from their customers,
ponents to uncertain Internet nisms and procedures supporting a provider to of which several are basing their business on
traffic demands. achieve such goals. This becomes more impor- an operational network.
Ola Espvik, Editor in Chief tant as the number of users and services grows.
Telektronikk 2/3.2001 1
• Increasing number of actors involved in ser- then treated in a set of articles in the second sec-
vice provision and delivery; connecting sys- tion, called traffic, routing, resources. Interdo-
tems and networks managed by different main, SLA, policyandmanagementis the fol-
actors, partly co-operating and partly compet- lowing section, addressing essential questions
ing, require adequate sets of means for a given for commercially offering services – in a fast,
actor to ensure its business and service levels accurate and automatic way. The papers deal
towards its customers. This is further compli- with internal procedures and systems (e.g. man-
cated by the dynamic commercial and techni- agement systems) as well as relations with other
cal environment that an actor faces. actors (e.g. agreements). Measurementsare piv-
otal to follow and document the performance of
Several aspects have to be addressed as part of the network. A set of papers is showing how to
Traffic Engineering. A selection of the topics carry out measurements and factors to consider.
has been included in this issue of Telektronikk. The last section is called systems and services,
These are divided into a number of sections as presenting a few areas where solutions for IP
shown in the table of contents. Firstly, a set of and Traffic Engineering are used or required,
papers of introductorynature presents an over- like for mobile, for optics, and for voice.
view of the IP suite, history of Internet and prin-
ciples of Traffic Engineering. Basic topics of As the use of abbreviationsflourishes, a com-
designing and operating an IP-based network are mon list is collected at the end of this issue.
2 Telektronikk 2/3.2001
Computers and Communication
Early Development of Computing and Internet-technology
– a Groundbreaking Part of Technical History
YNGVAR LUNDH
Building the Arpanet and using it as a laboratory for development was a major contribution both to
computing and to telecom. During the 1970s a very innovative and thorough research and development
took place under the leadership of the United States Department of Defense, Advanced Research
Projects Agency. From the outset the effort can best be characterized as basic technical research.
Tenresearch groups worked together as a team at developing the Internet-technology itself. At the same
time the development was influenced by many other research groups, mainly in the United States, who
actively pursued specialized networking applications, thereby creating needs and uncovering possi-
bilities. Concurrently dramatic developments took place both in circuit and device techniques for com-
puters and in the politics of telecom operation. A glimpse of the Norwegian perspective is included
through the eyes of the small participating group in Norway. The text endeavours to be readable without
Yngvar Lundh (69) graduated
from the Norwegian Institute of prior technical special knowledge. Its objective is to interrelate main technical events of computing in a
Technology in 1956. He served historic perspective.
as leader of development teams
and projects in computer and
telecom technology, with the
Norwegian Defence research Strengthening Telecom and World Wide Web – WWW
Establishment until 1984, then Computer Techniques People today tend to think of Internet as synony-
with the Norwegian Telecom
Administration / Telenor until Internet is not only a modern hobby, nor an mous with World Wide Web. That technique
1996, then in his own consulting exotic new trading arena, nor new mail, nor new opens a splendid new view of a world of infor-
company. Since 1980 he also
distribution of information. Internet is all of that mation. Simple point and click movements catch
served as Professor of Informat-
ics at the University of Oslo. His and much more. A result of a successful develop- posted information anywhere in the world inde-
main projects concerned digital ment since the late 1960s of basic new technical pendently of distance. Information is posted in
computing and circuit technol-
principles, it pertains to computer collaboration the form of “home pages”. The information is
ogy, enabling the start of “hi-
tech” companies, notably Norsk and to various forms of information transfer. We coded in standardized format and is stored in
Data AS, and systems for de- shall refer to these techniques as Internet-tech- computers connected to the network. For the
fence and for telecom enhanc-
nology. During the same period – the last 25 coding and formatting various aids are readily
ing and employing emerging
technologies. Lundh is a mem- years of the twentieth century – important devel- available and easy to use. Hence the “Web tech-
ber of he Norwegian Academy opment took place in two other areas. Electronic nology” can be used by anyone who wants a
of Technological Sciences.
circuit and device techniquesmade it possible message to be presented. It immediately be-
[email protected] tofabricate computers cheap and small, thereby comes available to the whole world. That is a
making them more interesting economically. world that can be described as “countless mil-
They can now be used as components in new lions” and which grows such that it will shortly
roles (for that matter the same circuit technology comprise at least everyone who has or could
also made even more powerful computers possi- have a telephone. At the same time the technol-
ble). Telecommunicationswent through a pro- ogy is developed further towards user-facilities
found reorganization. This admitted new actors with far more comprehensive abilities than tele-
and driving forces into the technical and com- phones at conveying information.
mercial evolution of telecom networks.
However, World Wide Web is far from the
Those are features of a development presently entire Internet. It is only one – admittedly very
causing drastic changes in our relationship with conspicuous – example of exciting new possibil-
information and its use. Most likely we are just ities that can be implemented on top of the Inter-
at the beginning of a new era. Nobody knows net-technology itself. The Web technique first
where it will take us. This technology is already emerged as a practical internal information dis-
beginning to change important functions of our tribution system in a large international research
society. Some of the perspectives that can now establishment in Switzerland – CERN – in
be perceived imply exciting new possibilities. 1990–91. From there it spread incredibly fast
throughout the Internet, i.e. to “the whole world”.
This article describes that technical development
without assuming special technical knowledge But the basic ideas were already 25 years old
and avoids detail that is better described else- then. The idea of being able to open windows
where. Outlines of the historic development of via CRT-screens on collections of information
the internetworking technology and parts of ofvarious types near or far and to navigate in a
computer- and telecom techniques are described. world of information by point and click move-
Special focus is directed at the 1970s when most ments of hands and fingers were demonstrated
of the basic technical development took place. already in 1968.
Telektronikk 2/3.2001 3
-----In any picture or text hyperlinks
may be built in. As an example, clicking stations were another of Engelbart’s ideas long
the name of a particular personmay before the microprocessor). And, of special
immediately bring in his importance, the ban on commercial traffic in the
picture located in some
Internet was lifted in 1991. Some of the prereq-
computer – anywhere in the
world uisites were met relatively early. Computer
screens, first demonstrated in the late 1950s,
• CV began to be common from the early 1970s, per-
• e-mail sonal computers a little later. Computer co-oper-
• web-add ation in general, vendor independent networks,
• work beyond the groups participating in developing
• home the Internet techniques themselves, became
usual from around 1980. Some of the ideas that
Engelbart demonstrated in 1968 are still (2001)
Mr. Ola Norman awaiting comprehensive use, but are expected
tobecome similarly important. Examples are
telephony and moving pictures.
Hyperlinks make information The leading person and creating and driving World Wide Web is primarily an example of
stored elsewhere available by force in these developments was Douglas Engel- technical possibilities that Internet technology
pointing at a reference to it bart of Stanford Research International – SRI – opens up as a carrier of entirely new and exciting
and clicking in Menlo Park, California. Engelbart developed forms of information handling.
and demonstrated a comprehensive set of con-
cepts and techniques of groundbreaking impor- Electronic Mail
tance. The most significant was perhaps the Message transfer was a dominating application
“mouse”, the little thing you hold in your hand. of the Internet already from the start of the
Everybody who has seen a computer today has development in 1969. That form of communica-
seen it. Equally important is the “hyperlink”. tion was especially practical and a necessary tool
That is a reference pointer-code that can be in the decentralized collaboration of the ten
embedded in any information-image – text or groups of researchers who undertook the basic
picture – displayed on a computer screen. A research and development in the 1970s.
mouse-click onthe pointer opens the referred
The concept of workstation image. That happens independently of where in The original vision of resource sharing network-
was demonstrated at SRI by the net that information happens to be located. ing, an important source of inspiration for the
Douglas Engelbart as early development, comprised a number of other,
as1968. It had cathode ray A quarter of a century would elapse before these more or less exotic, applications. E-mail was an
screen, a “mouse” for the ideas could be seriously employed and put to overwhelming generator of traffic for many
right hand and a five-finger extensive use. Only then the technical and eco- years. It perhaps still is, at least was so until the
key-set for the left hand. The nomic prerequisites were met. Microprocessors Web started another “landslide” of new users.
user could point, click and type had made low priced personal computers com- Tomany people either e-mail or Web is still
anywhere at the screen while mon as “workstations” and the Internet was synonymous with the Internet.
looking steadily available and ubiquitous (networking work-
E-mail is an important form of communication
already. It has unique properties in comparison
to ordinary mail and telephone. Therefore it
defends a place of its own as a communications
medium. It overcomes both time and distance,
itis fast, but still the addressee may answer pre-
cisely, for record, when convenient after having
had plenty of time to think, unlike the telephone.
E-mail is suitable for automation in various
forms. And it is cheap.
However, e-mail is still – in 2001 – far from
having reached its full potential as a general
communication medium. “Old-fashioned” mail
and telephone are far ahead in general availabil-
ity. The most important shortcoming is that e-
mail only reaches those who often use comput-
ers, connected to the Internet – and whenthey
use the computer. To make certain that a mes-
sage gets there, at least to most people, it will
still be best to call or send a letter.
4 Telektronikk 2/3.2001
E-mail will be much more usable the day when The Internet is a network of
Local
the receiving apparatus immediately emits a nets interconnected by gateway
area
beep or blinking light signalling the arrival of computers. Nets may be of
net
amessage. The messaging apparatus should be different types. From one host
G
equally cheap to own as a telephone that rarely computer to another
rings. We are still some distance from being able G Net of information travels in packets
leased
to send important messages to many recipients along routes which may
lines
by e-mail only. But the trend is pointing in that G change with network “shape”
direction. And a transmitter and receiver of e- Packet Packet and traffic load. The Internet
mail will not be a PC only. It will be built into radio G satellite Protocol (IP) helps navigate
telephones or other future “popular boxes”, that net net through the network. The
G
we have not yet seen, but which will creatively Transport Control Protocol
be made usual in the future world. Many such G G (TCP) helps ensure that a
boxes will be gadgets with remotely controlled message transported
functions, etc. Local Local piecemeal as a number of
area area packets gets properly
Wireless or Wired Somehow net net re-assembled
The actual reason for the name Internet is that
the network may employ various carrier media
Host A Host B
of many different kinds interconnected for the
transportation of information.
Further, Internet-technology implies mechanisms type of net. The individual nets are connected
for optimized mixing of different transport re- into one network (of nets) called Internet. Gate-
quirements. Urgent messages get there fast while way computers make the interconnection. They
less urgent traffic may be transported more cheap- convert the information on its way to the next
ly using otherwise idle periods. Traffic types are net into a format suitable for appropriate han-
more or less error prone. Particularly important dling there.
traffic needs precedence – e.g. for resolving
problems in the network itself, and so on. The basic development of Internet-technology
took place in the period 1969 – 1980. It com-
Characteristics such as error density, urgency, prised understanding the problems and the possi-
and precedence are measurable quantities to be bilities, creating and testing the technical meth-
specified, and to be met accordingly. Internet ods and defining the results as standards that
technology lets different requirements be met were open and available for use by anyone.
automatically by the available network capabilities.
The main result of this development is that dif-
In particular this enables many carrier media of ferent computers can now co-operate and ex-
rather different capabilities to co-operate in the change all types of information. Further, the
transport such that each medium is exploited to
its best ability. Prevailing media now are leased
lines capable of specific bit transfer rates – num-
ber of pulses per second. Various standard band-
widths (pulses per second) are available. Several
other carrier media are important and will be
used increasingly. Local area networks prevail
within buildings and geographically limited cor-
porate sites. Radio computer networking of vari-
ous kinds is evolving further.
Satellites have many exciting possibilities. The
same is true of cable networks originally built
for broadcasting television. These are probably
best suited for rural and urban areas respectively,
and have great potential for being exploited further.
The background for the name Internet is net of
interconnected nets. The individual transport Teleprinter, (“Teletype”,
networks are operated separately as mask-shaped “Telex machine”) built for the
nets of leased lines, packet radio nets, packet international Telex network
satellite nets, and so on. Each of these media was extensively used as
transports packets in ways best suited for each computer terminals
Telektronikk 2/3.2001 5
Remote Computers and
Time Sharing
Computer Early computers and their users communicated
center
using teleprinters – “Teletype machines” – that
conveyed text directly both into and out of the
computers. For many years information was
transferred in and out of computers via punched
Modems cards or punched paper tape. The need for effi-
cient use of computer time dictated these media.
They were much faster and hence occupied less
of the valuable computer time waiting for slow
Tele-net
fingers and printer mechanisms. Special line
printers to be directly driven by the computer
were developed. For many years powerful line-
printer machines were essential parts of com-
Modems puter centres, and they produced vast amounts
• • • • • • • • Terminals ofpaper printouts. Although they are capable of
fast printing of text on paper lineprinters have
been surpassed in performance by newer de-
vices. Both ink jet printers and xerographic
printing mechanisms using lasers for pattern
generation are almost household items today,
and they outperform the earlier, powerful mech-
anical line printers both at efficiency, quality and
Timesharing allows many information can be transported through different flexibility.
users to work directly and types of transport media. Each transport is auto-
simultaneously with the com- matically handled as required by interconnected The first operating systems were developed early
puter from remote locations carrier media that may have different character- in the 1960s. That is programs that manage the
istics. computer itself and its attached resources. Since
then a computer without an operating system is
Figure 5 CRT-screens and The transport is handled according to technical unthinkable. The operating system manages the
timesharing were major steps rules called protocols. They depend on more computer and its various tasks. As an example
in adapting the power of detail and are significantly more complicated the operating system permits the fast central
computers to the abilities of than the protocol that govern ordinary telephone processing unit to carry on at full speed while
people. This picture was copied traffic. In reality the protocols are implemented slower attached units such as printers work at
from a brochure in 1971. The as programs in computers. Microprocessors of theirspeed. Important operating systems today
large company impressed various processing powers are now available and are Windows, NT, UNIX and Linux, each in
customers calling in. As soon can be used as components in devices built to several variations. While Internet-technology is
as the customer gave his name use the Internet for various purposes. It appears independent of individual machines and operat-
the operator could (actually that we are presently at the beginning of an era ing systems, the one operating system that was
type it immediately and thus of future types of information networks. most prevalent during the first decade of net-
display his data, hence –) working development was “Tenex” by Digital
“remember” details about him Equipment Corporation – DEC. It was a popular
operating system especially for that company’s
tenth major computer model (“Programmed
Data Processor”) – PDP 10. Towards the end of
the 1970s Unix became more and more popular
and has become an important industry standard
serving many types of computers. In the 1990s
Microsoft Corporation’s various “Windows”
systems have overtaken it in volume outnumber-
ing all others.
A remarkable phenomenon during the period
ofinternetworking development is the absence
of IBM from that development. From the late
1950s that large, internationally distributed com-
pany had a uniquely large market share for com-
puters and everything in computing. Especially
IBM took leadership in many areas of technical
standard setting. Cards, magnetic tapes, codes,
etc. were “IBM-compatible”. IBM also set their
6 Telektronikk 2/3.2001
own comprehensive and successful standards in
computer networking.
The operating system enables a computer to do
several tasks at the same time. A particularly
important development was timesharing. The
Computer Time-Sharing System – CTSS – was
demonstrated in its earliest form at MIT in the
early to mid 1960s. Timesharing permits several
users to access the computer simultaneously.
Several user terminals, typically Teletype
machines, may be connected to the computer.
Each user experiences the communication with
the computer – via the timeshared operating sys-
tem – as if she had the computer to herself alone.
The computer actually shares its time between
several users and several tasks. The users per-
ceive increased load from more users as slower
response from the computer. Powerful comput-
ers may serve many, perhaps several tens of
users without noticeably slow response.
Standard Teletype machines are specialized
typewriting machines made to be connected for mode”. The actual users, i.e. those who devel- Special printer terminals
transfer of written – teletype – messages through oped programs and those who delivered data for became available
the international Telexnetwork. In the 1960s processing, were not allowed to communicate
specialized typewriting computer terminals directly with the computer. The user submitted
began to replace the use of Teletype machines jobs consisting of programs and/or data, usually
for computer purposes. They were faster and in the form of a deck of punched cards or a roll The line printer was the output
more flexible to use, had more comprehensive of punched paper tape. A popular profession was medium from computer
character sets and could produce nicer print. that of punching machine operator, transferring centers. It produced large
Only from the beginning of the 1970s terminals written programs or data from paper into amounts of printed pages
with Cathode Ray Terminal screens became punched cards, using the specialized “off-line” folded together
more and more usual. From around 1980 CRTs
were dominating computer use. Terminals are
connected to computers via [telephone] lines.
Gradually it became usual to employ the tele-
phone network for communication with comput-
ers. The signals between the terminal and the
computer were converted into signals that could
be transmitted similarly to speech signals. They
were “modulated” and “demodulated” by
modems. In this way users of computers could
have terminals placed at the users’ premises,
connected to computers elsewhere via fixed –
leased – or “dial-up” telephone lines.
Vendors of computing services established large
computer centres sometimes with extensive nets
of leased lines and connection points for modem
connections. The one single application making
most use of early computer networking was seat
reservation for airline passenger traffic. Already
in the 1960s the large airlines had ubiquitous,
often global networks for that purpose.
Computer Centres and
Personal Computers
For economic management of expensive com-
puters stable full employment of the machine is
important. Before the invention of timesharing
computer centres were run in “closed shop
Telektronikk 2/3.2001 7
From the mid 1970s that technical development
permitted complete programmable computers to
be made cheaply enough to make it meaningful
economically for one person to have the com-
puter alone. The movement of the human per-
son’s fingers and her ability to think and formu-
late commands and questions are very slow if
measured in a computer’s time scale. Looking at
the exploitation of a personal computer – PC –
one will typically find the computer running idle
most of the time while the user thinks or does
something else. The PC-phenomenon – econom-
ically seen – is that it makes economic sense to
have the machine idling, but immediately avail-
able to the user – the person. Unlike the situation
today that was far from true for many years.
Resource Sharing Networks
Computers began to have practical significance
in business as tools and production machinery
from the last half of the 1950s. Computer tech-
nology and its use have continued to improve
and increase since then. This technology has an
unbelievably large and increasing importance.
The punched card was an punching machines. Operating staff accepted The development has continued and continues
important input medium well jobs through a window and delivered results in on. All the time it becomes more specialized
into the 1970s. Itwas only the form of line printer print-outs (and the andrefined.
replaced when timesharing returned card decks). During program develop-
and sufficient disk storage ment that typically meant lists of error messages. Academic interest in this dramatic development
permitted users to enter their The primary task of the operating staff was to started in a few places. From the late 1950s it
programs directly and leave keep the computer in stable and reliable opera- was the basis of a new industry, rapidly growing.
them in the computer. One tion for high productivity, referring to the use of IBM was the foremost, gigantic, standard setting
card represented upto 80 computer time as an overwhelming cost factor. “powerhouse” among several great companies.
characters. (Pictures copied Those card decks could be large, sometimes fill- A large and growing number of engineers and
from “Britannica.com”) ing long steel drawers almost too heavy to carry. technical scientists engaged themselves in the
study and development of the great new poten-
Electronic integrated circuits and devices for tial that they saw and that began to open up
storage and presentation (semiconductor mem- around 1960.
ory, storage disks and CRT screens) have gone
through a very comprehensive development One broadly based research program started in
indeed continuously since 1960. Then the first the late 1960s. The Advanced Research Projects
digital circuits could be built in large scale using Agency – ARPA – of the United States Depart-
transistors rather than vacuum tubes. That devel- ment of Defense sponsored the building of a
opment improved performance characteristics resource-sharing network called Arpanet. It
bymany orders of magnitude (and continues connected four computers at universities and
rapidly to improve further). research establishments in the western United
States. The research program was basic technical
research and the visions about Arpanet were
resource sharing. Important and valuable re-
sources that could be shared and hence be better
exploited were several: Powerful computers
By 1970 the integrated circuit (although quite weak by today’s measures) were
technology had developed to a too expensive to be procured by all those who
point where microprocessors wished for them and who could have put them
could be “stamped out” togood use. It became desirable to make such
cheaply on silicon wafers in computer resources available to more groups
large numbers. This picture ofpeople. Thereby new problem areas might be
shows an experimental wafer attacked and manifold creativity might more eas-
with anarray of several ily meet in fruitful collaboration – resource shar-
different circuit chips. The ing.
matrix is cutinto individual
chips
8 Telektronikk 2/3.2001