NETWORK COMMUNICATION CONCEPTS
In this section we introduce concepts required to understand the design and operation of networks. We shall examine client-server architecture and distinguish between thin and fat clients. We then consider network topologies that describe how network devices are physically and logically connected. Finally we describe different encoding and decoding methods used to represent data as signals suitable for transmission.
CLIENT-SERVER ARCHITECTURE
As the name client-server suggests, there are two different types of computer present on the network, namely servers and clients. The server provides particular processing resources and services to each client machine. For example, web servers retrieve and transmit web pages, and database servers retrieve and transmit records. The client machines, which are commonly personal computers, also perform their own local processing. For example, web browsers, email clients and database applications. Each server provides processing services to multiple clients. Client-server processing is a form of distributed processing where different computers are used to perform the specific information processes necessary to achieve the systems purpose. Client-server processing occurs sequentially, this means that for each particular client-server operation just one CPU is ever processing data at a particular time. Many operations may well be occurring simultaneously however each particular operation is processed sequentially. When a particular operation is being performed either the client is processing or the server is processing, but not both at the same time.
The client machine performs processing and then when it requires the resources of the server it sends a request, the client waits for a response from the server before it continues processing. Between the request being sent and the response being received the server is performing the requested processes. Notice that the client machines do not need to understand the detail of the server’s processes and the server does not need to understand the detail of the processes occurring on the clients. Rather the two machines merely agree on the organisation of requests and responses. Hence a single server can provide processing resources to a variety of different clients running quite different software. For example, a single web server is able to provide resources to client computers of various types running a variety of different web browsers. Similarly a single database server can provide data to a variety of different client applications. As long as the request is legitimate, the server will perform the required processes and generate and transmit a response. Our discussion so far implies that servers are quite separate computers dedicated solely to server tasks; for large systems with many clients this is often the case, Cash dispensers include a safe that contains drawers for each denomination of bank note and another drawer for reject bills. The cash dispenser includes two sensors andvarious mechanical parts for moving bank notes. Onesensor counts the number of bills and the other measuresthe thickness of each bill. Any bills that do not meetspecifications are diverted to the reject drawer at the top of the safe.
Consider a small office or even home local area network (LAN). One machine is likely to be connected to the Internet and hence is anInternet server for all other computers on the LAN. Another computer on the LAN isconnected to and controls the operation of a shared printer; hence it is a print server.Both these computers are servers, yet they are also clients to each other and even tothemselves. In effect a computer can be a server for some tasks and a client for others.In general client applications provide the user interface, hence they manage allinteractions with end-users. This includes collecting and displaying information processes. In many cases the user is unaware of the server’s role – indeed many usersmaybe unaware of the servers very existence. From the user’s perspective interactions between client and server are transparent. For example when performing an Internet banking transaction a web browser is the client application that requests data from the banks web server. The banks web server then acts as a client to the banks DBMSserver. Users need not be aware of the servers involved and almost certainly areunaware of the specifics of the client-server processes occurring.On larger local area networks (LANs) it iscommon for all network tasks to be performed by one or more servers usingclient-server architecture. These serverscommonly run a network operating system(NOS) such as versions of Linux, Novell Netware or Window’s Server. These network servers control authentication of usersto ensure security. Authentication processes aim to determine if users, and other devices, are who they claim to be. Commonly users must log into the network server before they are able to perform any processing. In most cases a logon password isused, however digital certificates and biometric data such as fingerprints are becoming popular methods of authenticating users. NOSs also provide file server, print server and numerous other services to users. We examine NOSs and their capabilities in more detail later in this chapter.In our above discussion, the client machine has applications installed that are executed by the CPU within the machine. Such clients are known as “fat clients” or “thick clients”. Another strategy that is gaining in popularity is the use of thin clients. A thinclient is similar in many ways to the old terminals that once connected to centralisedmainframe computers. These terminals only performed basic processing tasks, such asreceiving data, displaying it on the screen and also transmitting input back to themainframe. Thin clients can be implemented in a number of ways. They can be very basic low specification personal computers, often without any secondary storage.These thin clients rely on servers to perform all the real processing. Other thin clientimplementations are software based. For instance, the RDP (Remote DesktopProtocol) can be used to connect and execute any application running on a remoteserver. Essentially RDP simply sends the screen display from the remote computer tothe thin client. The user at the thin client can therefore log into and operate the remotecomputer as if they were actually there. This technique is popular with IT staff as itallows them to manage servers from remote locations, such as from home. It is alsoroutinely used to allow employees to access their work network from home or other locations via the Internet. RDP and other thin client protocols also provide a simple technique for making applications available over the Internet.
Information Processes and Technology – The HSC Course
NETWORK TOPOLOGIES
The topology of a network describes the way in which the devices (nodes) areconnected. A node is any device that is connected to the network, includingcomputers, printers, hubs, switches and routers. All nodes must be able tocommunicate using the suite of protocols defined for the particular network. Ingeneral all nodes are able to both receive and transmit using the defined network protocols. Nodes are connected to eachother via transmission media, either wiredcable or wireless.The topology of a network describes theseconnections in terms of their physicallayout and also in terms of how data islogically transferred between nodes. The physical connections between devicesdetermine the physical topology. Thelogical topology describes how nodescommunicate with each other rather thanhow they are physically connected.There are three basic topologies – bus, star and ring. In addition two other topologies,hybrid and mesh, are common on larger networks. Each of these topologies candescribe the physical or the logical topology of a network. Often the logical topologyis different to the physical topology. For example a physical star topology has allnodes on the LAN connected by individual cables back to a central node – often a hubor switch. This same network can have a different logical topology, either a logical bus or perhaps a logical ring topology.
Physical Topologies
•Physical Bus Topology
All nodes are connected to a single backbone – also known as a trunk or bus. The backbone is a single cable that carries data packets to all nodes. Each node attachesand listens for data present on the backbone via a T-connector or vampire connector.As the two ends of the backbone cable are not joined it is necessary to installterminators at each end. The function of the terminators is to prevent reflection of thedata signal back down the cable. On electrical networks, as opposed to fibre opticnetworks, terminators are resistors that completely stop the flow of electricity byconverting it into heat.In the past physical bus topologies were used for most LANs – in particular Thick net and Thin net Ethernet LANs that use coaxial cable as the transmission media.Although these networks require less cable than current star wired topologies they are unable to accommodate the large number of nodes present on many of today’s LANs
Furthermore a single break in the backbone disables the entire network. Today physical bus topologies are used for some high-speed backbones (often using fibre optic cable) and other long distance connections within commercial and government Wans. These high-speed applications have few attached nodes, in many cases just one at each end of the backbone to link two buildings. Where quality of service is critical it is common to install a secondary backbone to provide a redundant connection. If the primary backbone fails for any reasons then the network automatically switches to the secondary backbone.
•Physical Star Topology
All nodes connect to a central node via their own dedicated cable. Today the physical star topology is used on almost all LANs, including wireless LANs. In most cases the central node is a switch that includes multiple ports. In the past the central node was likely to have been a hub, multi station access unit (MAU) or even a central computer. We consider the operation of hubs and switches later in this chapter. MAUs are used in token ring networks so that a physical star topology can be used with token ring’s logical ring topology. For wireless LANs a WAP (Wireless Access Point) is used as the central node. In terms of physical star topologies the central node is the device that connects all outlying nodes such that they can transmit and receive packets to and from each other node. Physical star topologies have a number of advantages over physical bus and ring topologies. This is particularly true for LANs where nodes are physically close – such as within the same room or building. Firstly each node has its own cable and hence can be connected and disconnected without affecting any other nodes. Secondly new nodes can easily be added without first disabling the network. Finally identifying faults is simplified as single nodes can simply be disconnected from the central node in turn until the problem is resolved. There are however some disadvantages of physical stars. Significantly more cabling is required, however this cable is generally less expensive as it must only support transmission speeds sufficient for a single node. Today UTP (Unprotected Twisted Pair) is the most common transmission media. Also if a fault occurs in the central node then all connected nodes are also disabled
Physical Ring Topology
In a physical ring each node connects to exactly two other nodes. As a consequence the cable forms a complete ring. In general data packets circulate the ring in just one direction. This means each node receives data from one node and transmits to the other. If the cable is broken at any point then the entire network is disabled. Therefore removing a node or adding a new node requires the network to be stopped. Furthermore in most implementations each data packet is received and then retransmitted by each node, hence all nodes must be powered at all times if the network is to operate. For these reasons physical ring topologies are seldom used for LANs today.FDDI (Fibre Distributed Data Interface) and SONET (Synchronous Optical Network)networks are usually configured as physical rings and always operate as logical rings. FDDI can be used for LANs however it is more commonly used for longer distance high-speed connections. As the names suggest FDDI and SONET use optical fibre as the transmission media. FDDI is commonly used to connect an organisation’s buildings whilst SONET is used for much greater distances. Both protocols use two physical rings with data circulating in different directions on each ring. Distances between FDDI nodes should not exceed 30km while distances in excess of 100km are common for SONET. For long distance applications the second ring is maintained solely as a backup should a fault occur in the primary ring? In such cases it is preferable to physically route the cabling of each ring separately. The aim being to improve fault tolerance should a cable be broken at any single location. If the cables for both rings are within close proximity (like within the same trench) then chances are that both cables will be broken together. When FDDI is used within a building then both rings can be used for data transmission, which effectively doubles the speed of data transfer.
•Physical Hybrid Topology
Hybrid or tree topologies use a combination of connected bus, star and ring topologies. Commonly a physical bus topology forms the backbone, with multiple physical star topologies branching off this backbone.
The backbone is installed through each building (or room) with a star topology used to branch out to the final workstations – the topology resembles the trunk and branches of a tree. All hybrid topologies have a single transmission path between any two nodes. This is one reason the name ‘tree’ is used; consider the leaves on a tree, there is one and only one path from one leaf to another – the same is true for nodes in a physical hybrid or tree network
In this section we introduce concepts required to understand the design and operation of networks. We shall examine client-server architecture and distinguish between thin and fat clients. We then consider network topologies that describe how network devices are physically and logically connected. Finally we describe different encoding and decoding methods used to represent data as signals suitable for transmission.
CLIENT-SERVER ARCHITECTURE
As the name client-server suggests, there are two different types of computer present on the network, namely servers and clients. The server provides particular processing resources and services to each client machine. For example, web servers retrieve and transmit web pages, and database servers retrieve and transmit records. The client machines, which are commonly personal computers, also perform their own local processing. For example, web browsers, email clients and database applications. Each server provides processing services to multiple clients. Client-server processing is a form of distributed processing where different computers are used to perform the specific information processes necessary to achieve the systems purpose. Client-server processing occurs sequentially, this means that for each particular client-server operation just one CPU is ever processing data at a particular time. Many operations may well be occurring simultaneously however each particular operation is processed sequentially. When a particular operation is being performed either the client is processing or the server is processing, but not both at the same time.
The client machine performs processing and then when it requires the resources of the server it sends a request, the client waits for a response from the server before it continues processing. Between the request being sent and the response being received the server is performing the requested processes. Notice that the client machines do not need to understand the detail of the server’s processes and the server does not need to understand the detail of the processes occurring on the clients. Rather the two machines merely agree on the organisation of requests and responses. Hence a single server can provide processing resources to a variety of different clients running quite different software. For example, a single web server is able to provide resources to client computers of various types running a variety of different web browsers. Similarly a single database server can provide data to a variety of different client applications. As long as the request is legitimate, the server will perform the required processes and generate and transmit a response. Our discussion so far implies that servers are quite separate computers dedicated solely to server tasks; for large systems with many clients this is often the case, Cash dispensers include a safe that contains drawers for each denomination of bank note and another drawer for reject bills. The cash dispenser includes two sensors andvarious mechanical parts for moving bank notes. Onesensor counts the number of bills and the other measuresthe thickness of each bill. Any bills that do not meetspecifications are diverted to the reject drawer at the top of the safe.
Consider a small office or even home local area network (LAN). One machine is likely to be connected to the Internet and hence is anInternet server for all other computers on the LAN. Another computer on the LAN isconnected to and controls the operation of a shared printer; hence it is a print server.Both these computers are servers, yet they are also clients to each other and even tothemselves. In effect a computer can be a server for some tasks and a client for others.In general client applications provide the user interface, hence they manage allinteractions with end-users. This includes collecting and displaying information processes. In many cases the user is unaware of the server’s role – indeed many usersmaybe unaware of the servers very existence. From the user’s perspective interactions between client and server are transparent. For example when performing an Internet banking transaction a web browser is the client application that requests data from the banks web server. The banks web server then acts as a client to the banks DBMSserver. Users need not be aware of the servers involved and almost certainly areunaware of the specifics of the client-server processes occurring.On larger local area networks (LANs) it iscommon for all network tasks to be performed by one or more servers usingclient-server architecture. These serverscommonly run a network operating system(NOS) such as versions of Linux, Novell Netware or Window’s Server. These network servers control authentication of usersto ensure security. Authentication processes aim to determine if users, and other devices, are who they claim to be. Commonly users must log into the network server before they are able to perform any processing. In most cases a logon password isused, however digital certificates and biometric data such as fingerprints are becoming popular methods of authenticating users. NOSs also provide file server, print server and numerous other services to users. We examine NOSs and their capabilities in more detail later in this chapter.In our above discussion, the client machine has applications installed that are executed by the CPU within the machine. Such clients are known as “fat clients” or “thick clients”. Another strategy that is gaining in popularity is the use of thin clients. A thinclient is similar in many ways to the old terminals that once connected to centralisedmainframe computers. These terminals only performed basic processing tasks, such asreceiving data, displaying it on the screen and also transmitting input back to themainframe. Thin clients can be implemented in a number of ways. They can be very basic low specification personal computers, often without any secondary storage.These thin clients rely on servers to perform all the real processing. Other thin clientimplementations are software based. For instance, the RDP (Remote DesktopProtocol) can be used to connect and execute any application running on a remoteserver. Essentially RDP simply sends the screen display from the remote computer tothe thin client. The user at the thin client can therefore log into and operate the remotecomputer as if they were actually there. This technique is popular with IT staff as itallows them to manage servers from remote locations, such as from home. It is alsoroutinely used to allow employees to access their work network from home or other locations via the Internet. RDP and other thin client protocols also provide a simple technique for making applications available over the Internet.
Information Processes and Technology – The HSC Course
NETWORK TOPOLOGIES
The topology of a network describes the way in which the devices (nodes) areconnected. A node is any device that is connected to the network, includingcomputers, printers, hubs, switches and routers. All nodes must be able tocommunicate using the suite of protocols defined for the particular network. Ingeneral all nodes are able to both receive and transmit using the defined network protocols. Nodes are connected to eachother via transmission media, either wiredcable or wireless.The topology of a network describes theseconnections in terms of their physicallayout and also in terms of how data islogically transferred between nodes. The physical connections between devicesdetermine the physical topology. Thelogical topology describes how nodescommunicate with each other rather thanhow they are physically connected.There are three basic topologies – bus, star and ring. In addition two other topologies,hybrid and mesh, are common on larger networks. Each of these topologies candescribe the physical or the logical topology of a network. Often the logical topologyis different to the physical topology. For example a physical star topology has allnodes on the LAN connected by individual cables back to a central node – often a hubor switch. This same network can have a different logical topology, either a logical bus or perhaps a logical ring topology.
Physical Topologies
•Physical Bus Topology
All nodes are connected to a single backbone – also known as a trunk or bus. The backbone is a single cable that carries data packets to all nodes. Each node attachesand listens for data present on the backbone via a T-connector or vampire connector.As the two ends of the backbone cable are not joined it is necessary to installterminators at each end. The function of the terminators is to prevent reflection of thedata signal back down the cable. On electrical networks, as opposed to fibre opticnetworks, terminators are resistors that completely stop the flow of electricity byconverting it into heat.In the past physical bus topologies were used for most LANs – in particular Thick net and Thin net Ethernet LANs that use coaxial cable as the transmission media.Although these networks require less cable than current star wired topologies they are unable to accommodate the large number of nodes present on many of today’s LANs
Furthermore a single break in the backbone disables the entire network. Today physical bus topologies are used for some high-speed backbones (often using fibre optic cable) and other long distance connections within commercial and government Wans. These high-speed applications have few attached nodes, in many cases just one at each end of the backbone to link two buildings. Where quality of service is critical it is common to install a secondary backbone to provide a redundant connection. If the primary backbone fails for any reasons then the network automatically switches to the secondary backbone.
•Physical Star Topology
All nodes connect to a central node via their own dedicated cable. Today the physical star topology is used on almost all LANs, including wireless LANs. In most cases the central node is a switch that includes multiple ports. In the past the central node was likely to have been a hub, multi station access unit (MAU) or even a central computer. We consider the operation of hubs and switches later in this chapter. MAUs are used in token ring networks so that a physical star topology can be used with token ring’s logical ring topology. For wireless LANs a WAP (Wireless Access Point) is used as the central node. In terms of physical star topologies the central node is the device that connects all outlying nodes such that they can transmit and receive packets to and from each other node. Physical star topologies have a number of advantages over physical bus and ring topologies. This is particularly true for LANs where nodes are physically close – such as within the same room or building. Firstly each node has its own cable and hence can be connected and disconnected without affecting any other nodes. Secondly new nodes can easily be added without first disabling the network. Finally identifying faults is simplified as single nodes can simply be disconnected from the central node in turn until the problem is resolved. There are however some disadvantages of physical stars. Significantly more cabling is required, however this cable is generally less expensive as it must only support transmission speeds sufficient for a single node. Today UTP (Unprotected Twisted Pair) is the most common transmission media. Also if a fault occurs in the central node then all connected nodes are also disabled
Physical Ring Topology
In a physical ring each node connects to exactly two other nodes. As a consequence the cable forms a complete ring. In general data packets circulate the ring in just one direction. This means each node receives data from one node and transmits to the other. If the cable is broken at any point then the entire network is disabled. Therefore removing a node or adding a new node requires the network to be stopped. Furthermore in most implementations each data packet is received and then retransmitted by each node, hence all nodes must be powered at all times if the network is to operate. For these reasons physical ring topologies are seldom used for LANs today.FDDI (Fibre Distributed Data Interface) and SONET (Synchronous Optical Network)networks are usually configured as physical rings and always operate as logical rings. FDDI can be used for LANs however it is more commonly used for longer distance high-speed connections. As the names suggest FDDI and SONET use optical fibre as the transmission media. FDDI is commonly used to connect an organisation’s buildings whilst SONET is used for much greater distances. Both protocols use two physical rings with data circulating in different directions on each ring. Distances between FDDI nodes should not exceed 30km while distances in excess of 100km are common for SONET. For long distance applications the second ring is maintained solely as a backup should a fault occur in the primary ring? In such cases it is preferable to physically route the cabling of each ring separately. The aim being to improve fault tolerance should a cable be broken at any single location. If the cables for both rings are within close proximity (like within the same trench) then chances are that both cables will be broken together. When FDDI is used within a building then both rings can be used for data transmission, which effectively doubles the speed of data transfer.
•Physical Hybrid Topology
Hybrid or tree topologies use a combination of connected bus, star and ring topologies. Commonly a physical bus topology forms the backbone, with multiple physical star topologies branching off this backbone.
The backbone is installed through each building (or room) with a star topology used to branch out to the final workstations – the topology resembles the trunk and branches of a tree. All hybrid topologies have a single transmission path between any two nodes. This is one reason the name ‘tree’ is used; consider the leaves on a tree, there is one and only one path from one leaf to another – the same is true for nodes in a physical hybrid or tree network