Tải bản đầy đủ
8 UNI ( user- network interface), NNI ( network- network interface)and INI ( inter- network interface)

8 UNI ( user- network interface), NNI ( network- network interface)and INI ( inter- network interface)

Tải bản đầy đủ

Open systems interconnection (OSI)

Figure 1.5

11

UNI, NNI and INI type interfaces between end devices and data networks.

designed for UNI interface, it is often the case that one of the networks may be required to
act as DCE, while the other acts as DTE. Symmetry is achieved simply by allowing both ends
to assume either the DCE or the DTE role — as they see fit for a particular purpose. Some
physical NNI interfaces are truly symmetric.
The third main type of interface is called the INI (inter-network interface) or ICI (intercarrier interface). This is the type of interface used between networks under different ownership, i.e., those administered by different operators. Most INI interfaces are based upon
standard NNI interfaces. The main difference is that an INI is a ‘less trusted’ interface than
an NNI so that certain security and other precautions need to be made. An operator is likely
to accept signals sent from one subnetwork to another across an NNI for control or reconfiguration of one his subnetworks, but is less likely to allow third party operators to undertake
such control of his network by means of an INI. In a similar way, information received from
an INI (e.g., for performance management or accounting) may need to be treated with more
suspicion than equivalent information generated within another of the operator’s subnetworks
and conveyed by means of an NNI.

1.9 Open systems interconnection (OSI)
In the early days of computing, the different computer manufacturers developed widely diverse
hardware, operating systems and application software. The different strengths and weaknesses
of individual computer types made them more suited to some applications (i.e., uses) than
others. As a result, enterprise customers began to ‘collect’ different manufacturers’ hardware
for different departmental functions (e.g., for bookkeeping, personnel records, order-taking,
stock-keeping, etc.).
The business efficiency benefits of each departmental computer system quickly justified the
individual investments and brought quick economic payback. But the demands on computers
and computer manufacturers quickly moved on, as company IT (information technology)
departments sought to interconnect their various systems rather than have to manually re-type
output from one computer to become input for another. As a result, there was pressure to

12

The Internet, email, ebusiness and the worldwide web (www)

develop a standard means for representing computer information (called data) so that it could
be understood by any computer. Similarly, there was a need for a standard means of electronic
conveyance between systems. These were the first standards making up what we now refer to
as open systems interconnection (OSI) standards.
It is useful to assess some of the problems which have had to be overcome, for this gives
an invaluable insight into how a data network operates and the reasons for the apparently
bewildering complexity. In particular, we shall discuss the layered functions which make up
the OSI (open systems interconnection) model.
When we talk as humans, we conform to a strict etiquette of conversation without even
realising it. We make sure that the listener is in range before we start talking. We know who
we want to talk to and check it is the right person in front of us before we start talking.
We make sure they are awake, paying attention, listening to us, not talking to or looking at
someone else. We know which language they speak, or ask them in clear, slow language at
the start. We change language if necessary. We talk slowly and clearly and keep to a simple
vocabulary if necessary. While we talk, we watch their faces to check they have heard and
understood. We repeat things as necessary. We ask questions and we elaborate some points
to avoid misunderstanding. When we are finished we say ‘goodbye’ and turn away. We know
to ‘hang up’ the telephone afterwards (if necessary), thereby ensuring that the next caller can
reach us.
Computers and data networks are complex, because they are not capable of thinking for
themselves. Every situation which might possibly arise has to have been thought about and a
suitable action must be programmed into it in advance. Computers have no ‘common sense’
unless we programme it into them. If one computer tries to ‘talk’ to another, it needs to check
that the second computer is ‘listening’. It needs to check it is talking to the right piece of
equipment within the second computer. (We might send a command ‘shut down’, intending
that a given ‘window’ on the screen of the second computer should receive the command
and that the ‘window’ should thus close. But if instead the receiving computer directs the
command to the power supply, the whole PC would shut down instead.)
When a computer starts ‘talking’, it has to ‘speak’ in a ‘language’ which the listening
computer can understand, and must use an agreed set of alphabetic characters. When ‘talking’
it has to check that the listener has heard correctly and understood. And when talked to itself,
it may be appropriate to stop ‘talking’ for a while in order to concentrate on ‘listening’ or
to wait for a reply. Finally, when the communication session is over, it is proper formally to
close the conversation. The ‘listener’ need no longer pay attention, and the ‘talker’ may turn
attention to a third party.
The list of potential problems and situations to be considered by designers of data networks
is a long one. Here are a few examples:
• Different types of computer, using different operating systems and programming languages
wish to ‘talk’ to one another;
• Data information is to be shared by different types of application (e.g., bookkeeping and
order-taking programs), which use information records stored in different data formats;
• Different character representations are used by the different systems;
• It is not known whether the computer we wish to send data to or receive data from is
active and ready to communicate with us;
• To ‘reach’ the destination device we must ‘transit’ several intermediate networks of different types;
• There are many different physical, electrical and mechanical (i.e., plug/socket) interfaces.

Open systems interconnection (OSI)

13

The OSI model
The open systems interconnection (OSI) model, first formalised as a standard by ISO (International Organization for Standardization) in 1983 subdivided the various data communications
functions into seven interacting but independent layers. The idea was to create a modular
structure, allowing different standard functions to be combined in a flexible manner to allow
any two systems to communicate with one another. Although the model no longer covers all
the functions of data networks which have come to be needed, the idea of ‘layered’ protocols
and protocol stacks has come to be a cornerstone of modern data communications. It is thus
useful to explain the basics of the model and the jargon which it lays down.
To understand the OSI model, let us start with an analogy, drawn from a simple exchange
of ideas in the form of a dialogue between two people as illustrated in Figure 1.6. The speaker
has to convert his ideas into words; a translation may then be necessary into the grammar
and syntax of a foreign language which can be understood by the listener; the words are
then converted into sound by nerve signals and appropriate muscular responses in the mouth
and throat. The listener, meanwhile, is busy converting the sound back into the original idea.
While this is going on, the speaker needs to make sure in one way or another that the listener
has received the information, and has understood it. If there is a breakdown in any of these
activities, there can be no certainty that the original idea has been correctly conveyed between
the two parties.
Note that each function in our example is independent of every other function. It is not
necessary to repeat the language translation if the receiver did not hear the message — a request
(prompt) to replay a tape of the correctly translated message would be sufficient. The specialist
translator could be getting on with the next job as long as the less-skilled tape operator was
on hand. We thus have a layered series of functions. The idea starts at the top of the talker’s
stack of functions, and is converted by each function in the stack, until at the bottom it turns
up in a soundwave form. A reverse conversion stack, used by the listener, re-converts the
soundwaves back into the idea.
Each function in the protocol stack of the speaker has an exactly corresponding, or so-called
peer function in the protocol stack of the listener. The functions at the same layer in the two
stacks correspond to such an extent, that if we could conduct a direct peer-to-peer interaction
then we would actually be unaware of how the functions of the lower layers protocols had

Figure 1.6

A layered protocol model for simple conversation.

14

The Internet, email, ebusiness and the worldwide web (www)

Figure 1.7 The Open Systems Interconnection (OSI) model.

been undertaken. Let us, for example, replace layers 1 and 2 by using a telex machine instead.
The speaker still needs to think up the idea, correct the grammar and see to the language
translation, but now instead of being aimed at mouth muscles and soundwaves, finger muscles
and telex equipment do the rest (provided the listener also has a telex machine, of course!).
We cannot, however, simply replace only the speaker’s layer-1 function (the mouth), if we do
not carry out simultaneous peer protocol changes on the listener’s side because an ear cannot
pick up a telex message!
As long as the layers interact in a peer-to-peer manner, and as long as the interface between
the function of one layer and its immediate higher and lower layers is unaffected, then it is
unimportant how the function of that individual protocol layer is carried out. This is the
principle of the open systems interconnection (OSI) model and all layered data communications protocols.
The OSI model sub-divides the function of data communication into seven layered and
peer-to-peer sub-functions, as shown in Figure 1.7. Respectively from layer 7 to layer 1 these
are called; the application layer, the presentation layer, the session layer, the transport layer,
the network layer, the data link layer and the physical layer. Each layer of the OSI model
relies upon the service of the layer beneath it. Thus the transport layer (layer 4) relies upon
the network service which is provided by the stack of layers 1–3 beneath it. Similarly the
transport layer provides a transport service to the session layer, and so on. The functions of
the individual layers of the OSI model are defined more fully in ISO standards (ISO 7498),
and in ITU-T’s X.200 series of recommendations. In a nutshell, they are as follows.

Application layer (Layer 7)
This layer provides communications functions services to suit all conceivable types of data
transfer, control signals and responses between cooperating computers. A wide range of application layer protocols have been defined to accommodate all sorts of different computer
equipment types, activities, controls and other applications. These are usually defined in a
modular fashion, the simplest common functions being termed application service elements
(ASEs), which are sometimes grouped in specific functional combinations to form application entities (AEs) — standardised communications functions which may be directly integrated
into computer programs. These communications functions or protocols have the appearance

Open systems interconnection (OSI)

15

of computer programming commands (e.g., get, put, open, close etc.). The protocol sets out
how the command or action can be invoked by a given computer programme (or application)
and the sequence of actions which will result in the peer computer (i.e., the computer at the
remote end of the communication link). By standardising the protocol, we allow computers to
‘talk’ and ‘control’ one another without misuse or misinterpretation of requests or commands.

Presentation layer (Layer 6)
The presentation layer is responsible for making sure that the data format of the application
layer command is appropriate for the recipient. The presentation layer protocol tells the recipient in which language, syntax and character set the application layer command is in (in other
words, which particular application layer protocol is in use). If necessary, the presentation
layer can undertake a format conversion.
The binary digits (called bits) in which information is stored as data within computers are
usually grouped in 8-bit patterns called bytes. Computers use different codes (of either one or
two bytes in length) to represent the different alphanumeric characters. The most commonly
used standard codes are called ASCII (American standard code for information interchange),
unicode and EBCDIC (extended binary coded decimal interchange code). Standardisation of
the codes for representing alphanumeric characters was obviously one of the first fundamental
developments in allowing inter-computer communication.

Session layer (Layer 5)
A session between two computers is equivalent to a conversation between two humans, and
there are strict rules to be observed. When established for a session of communication, the
two devices at each end of the communication medium must conduct their ‘conversation’ in
an orderly manner. They must listen when spoken to, repeat as necessary, and answer questions properly. The session protocol regulates the ‘conversation’ and thus includes commands
such as start, suspend, resume and finish, but does not include the actual ‘content’ of the
communication.
The session protocol is rather like a tennis umpire. He or she cannot always tell how hard
the ball has been hit, or whether there is any spin on it, but he/she knows who has to hit the
ball next and whose turn it is to serve, and he/she can advise on the rules when there is an
error, in order that the game can continue. The session protocol negotiates for an appropriate
type of session to meet the communication need, and then it manages the session.
A session may be established between any two computer applications which need to communicate with one another. In this sense the application may be a ‘window’ on the computer
screen or an action or process being undertaken by a computer. Since more than one ‘window’
may be active at a time, or more than one ‘task’ may be running on the computer, it may
be that multiple ‘windows’ and ‘tasks’ are intercommunicating with one another by means of
different sessions. During such times, it is important that the various communications sessions
are not confused with one another, since all of them may be sharing the same communications
medium (i.e., all may be taking place on the same ‘line’).

Transport layer (Layer 4)
The transport service provided by the transport layer protocol provides for the end-to-end data
relaying service needed for a communication session. The transport layer itself establishes a
transport connection between the two end-user devices (e.g., ‘windows’ or ‘tasks’) by selecting

16

The Internet, email, ebusiness and the worldwide web (www)

Figure 1.8

Protocol multiplexing and splitting.

and setting up the network connection that best matches the session requirements in terms of
destination, quality of service, data unit size, flow control, and error correction needs. If more
than one network is available (e.g., leaseline, packet-switched network, telephone network,
router network, etc.), the transport layer chooses between them.
An important capability of the transport protocol is its ability to set up reliable connections
in cases even when multiple networks need to be traversed in succession (e.g., a connection
travels from LAN (local area network) via a wide area network to a second LAN). The
IP-related protocol TCP (transmission control protocol) is an example of a transport layer
protocol, and it is this single capability of TCP combined with IP (TCP/IP), that has made the
IP-suite of protocols so widely used and accepted.
The transport layer supplies the network addresses needed by the network layer for correct
delivery of the message. The network address may be unknown by the computer application
using the connection. The mapping function provided by the transport layer, in converting
transport addresses (provided by the session layer to identify the destination) into networkrecognisable addresses (e.g., telephone numbers) shows how independent the separate layers
can be: the conveyance medium could be changed and the session, presentation and application
protocols could be quite unaware of it.
The transport protocol is also capable of some important multiplexing and splitting functions
(Figure 1.8). In its multiplexing mode the transport protocol is capable of supporting a number
of different sessions over the same connection, rather like playing two games of tennis on the
same court. Humans would get confused about which ball to play, but the transport protocol
makes sure that computers do nothing of the kind.
Two sessions from a mainframe computer to a PC (personal computer) in a remote branch
site of a large shopping chain might be used simultaneously to control the building security
system and (separately) to communicate customer sales. Different software programmes (for
‘security’ and for ‘sales’) in both the mainframe computer and in the PC could thus share the
same telecommunications line without confusion.
Conversely, the splitting capability of the transport protocol allows (in theory) one session
to be conducted over a number of parallel network communication paths, like getting different
people to transport the various volumes of an encyclopaedia from one place to another.
The transport protocol also caters for the end-to-end conveyance, segmenting or concatenating (stringing together) the data as the network requires.

Network layer (Layer 3)
The network layer sets up and manages an end-to-end connection across a single real network,
determining which permutation of individual links need be used and ensuring the correct

EDI (electronic data interchange)

17

transfer of information across the single network (e.g., LAN or wide area network). Examples
of layer-3 network protocols are IP and X.25.

Datalink layer (Layer 2)
The datalink layer operates on an individual link or subnetwork part of a connection, managing the transmission of the data across a particular physical connection or subnetwork (e.g.,
LAN — local area network) so that the individual bits are conveyed over that link without
error. ISO’s standard datalink protocol, specified in ISO 3309, is called high level data link
control (HDLC). Its functions are to:
• synchronise the transmitter and receiver (i.e., the link end devices);
• control the flow of data bits;
• detect and correct errors of data caused in transmission;
• enable multiplexing of several logical channels over the same physical connection.
Typical commands used in datalink control protocols are thus ACK (acknowledge), EOT (end
of transmission), etc. Another example of a ‘link’ protocol is the IEEE 802.2 logical link
control (LLC) protocol used in Ethernet and Token Ring LANs (local area networks).

Physical layer (Layer 1)
The physical layer is concerned with the medium itself. It defines the precise electrical, interface and other aspects particular to the particular communications medium. Example physical
media in this regard are:
• the cable of a DTE/DCE interface as defined by EIA RS-232 or ITU-T recommendations:
X.21, V.35, V.36 or X.21bis (V.24/V.28);
• a 10 Mbit/s ethernet LAN based on twisted pair (the so-called 10baseT medium);
• a 4 Mbit/s or 16 Mbit/s Token ring LAN using Twinax (i.e., 2 x coaxial cable);
• a digital leaseline (e.g., conforming to ITU-T recommendation I.430 or G.703);
• a high speed digital connection conforming to one of the SONET (synchronous optical
network) or SDH (synchronous digital hierarchy) standards (e.g., STM-1, STM-4, STM-16,
OC3, OC12, STS3 etc.);
• a fibre optic cable;
• a radio link.

1.10 EDI (electronic data interchange)
By the 1980s, companies had managed to interconnect their different department computer
systems for book-keeping, order-taking, salaries, personnel, etc., and the focus of development
turned towards sharing computer data directly with both suppliers and customers. Why take an
order over the telephone when the customer can submit it directly by computer — eliminating
both the effort of taking down the order and the possibility of making a mistake in doing so?

18

The Internet, email, ebusiness and the worldwide web (www)

In particular, large retail organisations and the car manufacturers jumped on the bandwagon
of EDI (electronic data interchange).
The challenge of electronic data interchange (EDI) between different organisations is considerably greater than the difficulties of ‘mere’ interconnection of different computers as
originally addressed by OSI. When data is transferred only from one machine to another
within the same organisation, then that organisation may decide in isolation which information should be transferred, in which format and how the information should be interpreted
by the receiving machine. But when data is moved from one organisation to another, at least
three more problems arise in addition to those of interconnection:
• The content and meaning of the various information fields transferred must be standardised
(e.g., order number format and length, address fields, name fields, product codes and
names, etc.).
• There needs to be a means of reliable transfer from one computer to the other which
allows the sending computer to send its information independently of whether the receiving
computer is currently ready to receive it. In other words, the ‘network’ needs to cater for
store-and-retrieve communication between computers (comparable with having a postbox
at the post office for incoming mail which allows you to pick up your mail at a time
convenient to you as the receiver).
• There needs to be a way of confirming correct receipt.
Various new standardisation initiatives emerged to support EDI, among the first of which were:
• The standardisation of bar codes and unique product identification codes for a wide range
of grocery and other retail products was undertaken. The industry-wide standard codes
provided the basis for the ‘just-in-time’ re-stocking of supermarket and retail outlet shelves
on an almost daily basis by means of EDI.
• The major car manufacturers demanded EDI capability from their component suppliers,
so that they could benefit from lower stock levels and the associated cost benefits of
‘just-in-time’ ordering. Car products and components became standardised too.
• The banking industry developed EFTPOS (electronic funds transfer at the point-of-sale)
for ensuring that your credit card could be directly debited while you stood at the till.
All of the above are examples of EDI, and whole data networking companies emerged specialising in the needs of a particular industry sector, with a secure network serving the particular
‘community of interest’. Thus, for example, the ODETTE network provided for EDI between
European car manufacturers. TRADERNET was the EDI network for UK retailers. SWIFT is
the clearing network of the banks and SITA was the network organisation set up as a cooperative venture of the airlines for ticket reservations and flight operations. Subsequently, some
of these networks and companies have been subsumed into other organisations, but they were
important steps along the road to modern ebusiness (electronic business).
The store-and-retrieve methods used for EDI include email and the ITU’s message handling system (MHS) [as defined in ITU-T recommendation X.400]. Both are application layer
protocols which cater for the store-and-retrieve method of information transport, as well as
the confirmation of reply.

1.11 CompuServe, prestel, minitel, BTx (Bildschirmtext)
and teletex
The idea of equipping customers with computer terminals, so that they could log-in to a
company’s computers and make direct enquiries about the prices and availability of products

CompuServe, prestel, minitel, BTx (Bildschirmtext) and teletex

19

and services emerged in the later 1970s. Equipping the customer with the terminal improved
the level of customer service which could be offered, while simultaneously reducing the
manpower required for order-taking. Since the customer was unlikely to put a second terminal
on his desk (i.e., a competitor’s terminal), it also meant reduced competition.
The travel industry rapidly reorganised its order-taking procedures to encompass the use of
computer terminals by customers. There was soon a computer terminal at every airport check-in
desk and even some large travel agents. Other travel agents, meanwhile, continued to struggle
making phone calls to overloaded customer service agent centres. For a real revolution, all the
travel agents needed a terminal and an affordable means of network access. It came with the
launch of the first dial-up information service networks, which appeared in the late 1970s and
early 1980s. The first information services were the Prestel service of the British Post Office
(BPO) and the CompuServe information service in the USA (1979). Both were spurred by
the modem developments being made at the time by the Hayes company (the Hayes 300 bit/s
modem appeared in 1977).
The Prestel service followed the invention by the BPO laboratories of a simple terminal
device incorporating a modem and a keyboard, which could be used in conjunction with a
standard TV set as a ‘computer terminal’ screen. It spurred a new round of activity in the
ITU-T modem standardisation committees — as the V.21, V.22 and V.23 modems appeared.
And it became the impetus for the new range of teletex services which were to be standardised by ITU-T. The facsimile service appeared at almost the same time and also saw rapid
growth in popularity, so that the two together — teletex and facsimile tolled the death knell
for telex — the previous form of text and data communication which had developed from the
telegraph.
Other public telephone companies rapidly moved to introduce their own versions of teletex. France T´el´ecom introduced the world-renowned minitel service (Figure 1.9) in 1981 and
Germany’s Deutsche Bundespost introduced Bildschirmtext (later called BtX and T-Online

Figure 1.9

France T´el´ecom’s first minitel terminal (1981) [reproduced courtesy of France T´el´ecom].