NETWORK CONNECTION DEVICES
In this section we examine devices used to connect nodes to form a LAN and also to transfer data between networks. Each node requires a network interface card that complies with the Transmission Level protocols used by the network. For most Lana physical star topology is used hence a central node in the form of a hub, switch or wireless access point is required. Gateways connect networks that use different Transmission Level protocols whilst bridges connect networks using the same low-level protocols. Modems allow LANs to communicate with WANs. Routers operate at the Communication Control and Addressing Level to direct data along the most efficient path. For small LANs the functions of many of these devices is combined within a single hardware device generically known as a router.
•Network Interface Card (NIC)
Network interface cards convert data between the computers (commonly the PCI bus) into a form suitable for transmission across the network. The conversion uses the rules of the data link and physical link protocols in operation. It is the NIC that negotiates access to the network, including collision detection (or avoidance). Each NIC has its own unique MAC address so that other low-level network devices can uniquely identify the node. In the past most network interface cards were indeed cards that plugged into the motherboard. Today most computers include the functionality of an Ethernet NIC into the motherboard. An RJ45 port is included for connecting standard UTP patch cables. In addition most laptop computers include built in support for wireless LANs. Wireless NICs that connect via a USB or PCMCIA port are often used when the computer does not have an embedded wireless NIC. NICs for optical fibre networks are usually separate cards that install into a free slot on the PCI bus.
•Repeater
A repeater is any device that receives a signal, amplifies it and then transmits the amplified signal down another link. Repeaters are used to increase the physical range of the transmission media. Dedicated repeaters are routinely used to extend the reach of fibre optic cable. Most wireless access points can be used as simple repeaters to extend the coverage range of WLANs. Transponders used for ground-based and satellite microwave transmissions are also repeaters.
•Hub
When a hub receives a packet of data it simply amplifies and retransmits the packet to all attached nodes. As a consequence hubs are also known as multi-port repeaters. Hubs are dumb devices that operate at the physical layer of this model. They make no attempt to identify the destination node for each message. Hubs connect nodes together into a single network segment. This means all nodes attached to a central hub are sharing the same transmission channel meaning a logical bus topology is being Used. Hubs were once the primary devices used to connect UTP Ethernet networks. Today hubs have been largely phased out in favour of more intelligent switches.
•Bridge
A bridge separates a network into different segments at the data link layer. Bridges were once used extensively to segment Ethernet logical bus networks – today switches perform this function. Bridges determine the destinationMAC address of each frame. If the destination node with that MAC address is on the other side of the bridge then the frame is repeated onto that segment, otherwise the frame is dropped. Essentially a bridge splits a logical bus network into two collision domains.
•Switch
A switch can be thought of as an intelligent hub or amulet-port bridge. Switches determine the MAC address of the sender and intended receiver that precedes each message. The receiver’s address is used to identify the destination node and forward the message to that node only. In essence, a switch sets up a direct connection between the sender and the receiver; therefore each node exists on its own segment, the switch being the only other device on the segment. As no other nodes exist on each segment each node is free to transmit messages at any time without the need to detect or avoid collisions. Switches are able to simultaneously receive and forward messages from and to multiple pairs of nodes. As long as both the sender and the receiver of each message do not conflict with other simultaneous messages then the switch will direct the message correctly. Most switches allow nodes to communicate in full duplex.
•Gateway
A gateway connects two networks together. Gateways can connect networks that use different lower level protocols; however they can also be used to filter traffic movements between two similar networks. Gateways are routinely used to connect Alan to the Internet; however they can be used to connect any two networks. For example ADSL and cable modems (often called routers) include gateway functionality to convert between the low level Ethernet protocol used by the LAN and the low level protocols used by ADSL and cable connections. Larger LANs often include proxy servers whose task can include gateway functionality as they convert and filter traffic flowing between the LAN and the Internet. Gateways that connect IP LANs to the Internet have two IP addresses. A local address used for communication within the LAN and an Internet IP address used on the WAN or Internet side of the gateway. The local LAN IP address is used as the default
Gateway address for all local nodes wishing to access the Internet. The gateway hides the local IP addresses from the Internet; instead IP datagrams are all sent using the gateway’s WAN or Internet IP address. The gateway keeps track of the local I addresses so that IP traffic from the Internet can be directed to the correct local node. If a LAN includes a gateway that provides a connection to the Internet then the gateway’s LAN IP address must be known to all nodes – in most operating systems this IP address is specified as the default gateway
•Wireless Access Point
Wireless access points (WAPs) or simply access points (APs) are the central nodes on wireless LANs. Access points broadcast to all wireless nodes within the coverage area. On 802.11WLANs the access point does not direct packets to specific nodes or control the order in which nodes can transmit; rather they simply repeat all packets received. Conceptually an access point performs much like a hub on a wired LAN.A significant issue with WLANs is security – any user within the coverage range can potentially access the network. To counteract this possibility access points include security in the form of WEP (Wired Equivalent Privacy) and WPA (Wi-Fi Protected Access). WEP uses a single shared key encryption system whilst WPA generates new encryption keys at regular intervals. The WEP system can and has been infiltrated so currently WPA is the recommended system. No encryption system can work if it is not turned on. This is a major issue for both home and business WLANs. Furthermore the simplicity of creating a WLAN and the ability to access WLANs from outside make security a significant issue. Hackers need only to connect a wireless access point to an existing Ethernet connection point and they then have complete access without the need to work around complex firewalls and proxy servers.
Modem
The term modem is a shortened form of the terms modulation and demodulation, these are the primary processes performed by all modems. Today most modems are used to connect a computer to a local Internet Service Provider (ISP); the ISP supplying a high-speed ADSL or cable connection to the Internet. Dial-up modems were once the primary device for connecting users to the Internet. Currently dial-up modems are more often used to send faxes from computers over the PSTN – virtually all dial-up modems are able to both send and receive fax transmissions. We discussed modulation in some detail earlier in this chapter. Basically modems modulate digital signals by altering the phase, amplitude and/or frequency of electromagnetic waves. That is, modulation is the process of encoding digital data onto an analog waveform. Demodulation is the reverse of the modulation process. Demodulation decodes analog signals back into their original digital form. Clearly both sender and receiver must agree on the method of modulation used if communication is to be successful. Modems are commonly connected to a computer via a USB port or an Ethernet network connection. These interfaces are considered digital links; they do use electromagnetic waves however the data is represented using different voltages. The electronic circuits within the computer can use these voltage changes directly. In contrast modulated analog waves, such as those transmitted down telephone lines or coaxial cables, are not suitable for direct use by the circuits within the computer. Hence the primary role of modems is to provide an interface between the modulatedanalog waves used for long distance transfer and the digital data suitable for use by computers.
ADSL modems
Asymmetrical digital subscriber lines (ADSL) use existing copper telephone lines to transfer broadband signals. Although these copper wires were originally designed to support voice frequencies from 200 to 3400Hz, they are physically capable of supporting a much wider range of frequencies. It is the various switching and filtering hardware devices within the standard telephone network that prevent the transfer of frequencies above about 3400Hz. To solve this problem requires dedicated hardware to be installed where each copper line enters the local telephone exchange. ADSL signal strength deteriorates as distances increase; the signal cannot be maintained at all for distances greater than about 5400 metres. Voice lines much greater than 5400 metres are possible using amplifiers. Unfortunately these amplifiers boost only the lower frequencies required for voice, hence ADSL is not currently available in many remote rural areas. Even when distances are short and the copper runs directly into the exchange problems can occur as a consequence of interference. In general phone lines within a building and out to the street are not shielded against interference, this interference is rarely significant enough that a connection cannot be established; however it often reduces the speed of such connections.
So how does ADSL transfer data between an ADSL modem and the local telephone exchange? Using a modulation standard known as Discrete Multitude (DMT). DMT operates using frequencies from about 8kHz to around 1.5MHz.This bandwidth is split into some 247individual 4kHz wide channels . Each channel is modulated using QAM. DMT’s task is to specify the channels that are used for actual data transfer. If interference is present on a particular 4 kHz channel then DMT will shut down that channel and assign a new channel. This channel switching occurs in real time and is completely transparent to the user. In a sense ADSL is like having 247 dial-up modems all working together, each modem using QAM and DMT ensuring they all work together efficiently. The ADSL modem and the DSL hardware at the telephone exchange communicate to agree on the channels currently being used. At the local telephone exchange all the copper wires from the neighbourhood are connected to a splitter (see
Multiplexor
Simply refers tithe DSLAM’s task of combining multiple signals from customers onto single line and extracting individual customer signals from this single line. In most ADSL systems the lower bandwidth ADSL channels are used for upstream data (from modem to exchange) and higher frequency channels are used for downstream data (exchange to modem). Some channels are able to transfer data in both directions. ADSL is one example of a DSL technology, the A stands for asymmetrical, meaning transmitting and receiving occur at different speeds.
Cable modems
Cable modems connect to the Internet via coaxial cables; usually the same cable that transmits cable TV stations.
Fig 3.92
Describes how the bandwidth within the cable is split into channels. A single 6MHz bandwidth channel is used for downstream data – 6MHz is the width of a single cable TV station. This 6MHz wide channel is assigned within the range 88 to 860megahertz. A narrower bandwidth channel is used for upstream; commonly 1.6MHzwide however various other bandwidths are supported ranging from 200 kHz to3.2MHz. The upstream channel is assigned within the range 5 to 42 megahertz. The particular frequencies used for both channels are determined by the cable Internet provider and cannot be altered by individual users. The bandwidth used in a cable system is significantly larger than that used for ADSL.Therefore, one would assume the rate of data transfer would be much larger. In reality cable connections achieve speeds similar to ADSL connections; why is this? Cable connections are shared amongst multiple users. Single 6MHz downstream channels likely to be shared by hundreds of users. In a sense all the cable modems sharing a particular channel form a local area network. Every cable modem within the network receives all messages; they just ignore messages addressed to other modems. Consequently when only a few users are downloading then higher speeds are possible than when many users are downloading. Clearly the same situation occurs when uploading. This is why cable Internet companies include statements within their conditions stating that speeds quoted are not guaranteed
Router
Routers specialise in directing messages over the most efficient path to their destination. Today the large majority of routers operate at the network layer of this model using the IP protocol. Therefore routing decisions are based on each datagram’s destination IP address. Routers usually include the functionality of gateway. They are able to communicate with networks that use different protocols and even completely different methods and media for communication. Many routers also include variety of different security features. They are able to block messages based on the sender’s Padres, block access to specific web sites and even restrict communication to certain high-level protocols. Home or small business routers connect a singleLAN to the Internet. For these systems the decision is relatively simple – either the I datagram is addressed to a local node or it is not. Local datagrams are left alone whilst all others are sent out to the Internet. The routing table maintained by these routers is relatively small and rarely changes. Home and small business routers are commonly integrated devices that commonly include a router, an Ethernet switch and also wireless access point – these integrated devices are what the general public callrouters.Routers out on the larger Internet connects too many other routers. For these routers deciding on the best path for each IP datagram is considerably more complex. Such routers communicate with other adjoining routers to continually update their internal routing table. The routing table is examined to determine the most efficient route for each IP datagram. However, should any connections within the most efficient path fail then routers automatically direct the message over an alternate path? On larger wide area networks, and in particular the Internet, thousands of routers work together to pass messages to their final destination.
SERVERS
Servers provide specific processing services to other nodes (clients). We discussed the general operation of client-server architectures earlier in this chapter. In this section we briefly consider some of the more common services performed by servers. Note that this section is included under the general heading of “Network Hardware”; servers are often distinct computers designed with hardware suited to the services they provide, however what makes them servers is actually the installed software. On large networks dedicated servers are common whilst on smaller networks a server may well perform many tasks including the execution of end-user applications. Most servers run a network operating system (NOS) to manage user access to the services the server provides. We discuss features of network operating systems in the next section. Most network operating systems include file server and print server functionality as these are the core services that require user authentication and user access rights. There are numerous different services that servers provide. Examples of servers include file servers, print servers, database servers, mail servers, web servers and proxy servers. In this section we restrict our discussion to a brief overview of each of these services.
File Servers
A file server manages storage and retrieval of files and also application software in response to client requests. In hardware terms dedicated file servers do not require extremely fast processors, their main requirement being large amounts of fast secondary storage and a sufficiently fast connection to the network. Commonly file servers include multiple hard disks connected together into an array – RAID (Redundant Array of Independent Disks). Users are often unaware that multiple disks are being used. RAID uses different combinations of striping and mirroring to both improve data access speeds and also to improve the fault tolerance of the system. Striping stores single files across a number of physical disks and mirroring stores the same data on more than one disk. On larger RAID systems it is possible to replace faulty drives without halting the system – this is
Known as hot swapping. To further improve fault tolerance many file servers include various other redundant components including extra power supplies, cooling fans and in some cases the complete server is replicated. File servers must be able to process multiple file access requests from many users. Consequently the network connection to a file server often operates at a higher speed than for other workstation nodes. For each client request the file server, in combination with the NOS, checks the user’s access rights or permissions before retrieving the file. The file server in combination with the NOS ensures the file is retrieved and transmitted according to the user’s assigned access rights
Web Servers
We discussed the operation of web servers when discussing the HTTP protocol earlier in this chapter. Essentially a web server provides services to web browsers – they retrieve web pages and transmit them back to the requesting client web browser. Web servers must also include services that allow web pages to be uploaded, edited and deleted. Such services require users to first be authenticated by the web server. Many web servers, particularly those operated by ISPs, host many different web sites. These servers require high speed links to the Internet together with fast access to the files they host
In this section we examine devices used to connect nodes to form a LAN and also to transfer data between networks. Each node requires a network interface card that complies with the Transmission Level protocols used by the network. For most Lana physical star topology is used hence a central node in the form of a hub, switch or wireless access point is required. Gateways connect networks that use different Transmission Level protocols whilst bridges connect networks using the same low-level protocols. Modems allow LANs to communicate with WANs. Routers operate at the Communication Control and Addressing Level to direct data along the most efficient path. For small LANs the functions of many of these devices is combined within a single hardware device generically known as a router.
•Network Interface Card (NIC)
Network interface cards convert data between the computers (commonly the PCI bus) into a form suitable for transmission across the network. The conversion uses the rules of the data link and physical link protocols in operation. It is the NIC that negotiates access to the network, including collision detection (or avoidance). Each NIC has its own unique MAC address so that other low-level network devices can uniquely identify the node. In the past most network interface cards were indeed cards that plugged into the motherboard. Today most computers include the functionality of an Ethernet NIC into the motherboard. An RJ45 port is included for connecting standard UTP patch cables. In addition most laptop computers include built in support for wireless LANs. Wireless NICs that connect via a USB or PCMCIA port are often used when the computer does not have an embedded wireless NIC. NICs for optical fibre networks are usually separate cards that install into a free slot on the PCI bus.
•Repeater
A repeater is any device that receives a signal, amplifies it and then transmits the amplified signal down another link. Repeaters are used to increase the physical range of the transmission media. Dedicated repeaters are routinely used to extend the reach of fibre optic cable. Most wireless access points can be used as simple repeaters to extend the coverage range of WLANs. Transponders used for ground-based and satellite microwave transmissions are also repeaters.
•Hub
When a hub receives a packet of data it simply amplifies and retransmits the packet to all attached nodes. As a consequence hubs are also known as multi-port repeaters. Hubs are dumb devices that operate at the physical layer of this model. They make no attempt to identify the destination node for each message. Hubs connect nodes together into a single network segment. This means all nodes attached to a central hub are sharing the same transmission channel meaning a logical bus topology is being Used. Hubs were once the primary devices used to connect UTP Ethernet networks. Today hubs have been largely phased out in favour of more intelligent switches.
•Bridge
A bridge separates a network into different segments at the data link layer. Bridges were once used extensively to segment Ethernet logical bus networks – today switches perform this function. Bridges determine the destinationMAC address of each frame. If the destination node with that MAC address is on the other side of the bridge then the frame is repeated onto that segment, otherwise the frame is dropped. Essentially a bridge splits a logical bus network into two collision domains.
•Switch
A switch can be thought of as an intelligent hub or amulet-port bridge. Switches determine the MAC address of the sender and intended receiver that precedes each message. The receiver’s address is used to identify the destination node and forward the message to that node only. In essence, a switch sets up a direct connection between the sender and the receiver; therefore each node exists on its own segment, the switch being the only other device on the segment. As no other nodes exist on each segment each node is free to transmit messages at any time without the need to detect or avoid collisions. Switches are able to simultaneously receive and forward messages from and to multiple pairs of nodes. As long as both the sender and the receiver of each message do not conflict with other simultaneous messages then the switch will direct the message correctly. Most switches allow nodes to communicate in full duplex.
•Gateway
A gateway connects two networks together. Gateways can connect networks that use different lower level protocols; however they can also be used to filter traffic movements between two similar networks. Gateways are routinely used to connect Alan to the Internet; however they can be used to connect any two networks. For example ADSL and cable modems (often called routers) include gateway functionality to convert between the low level Ethernet protocol used by the LAN and the low level protocols used by ADSL and cable connections. Larger LANs often include proxy servers whose task can include gateway functionality as they convert and filter traffic flowing between the LAN and the Internet. Gateways that connect IP LANs to the Internet have two IP addresses. A local address used for communication within the LAN and an Internet IP address used on the WAN or Internet side of the gateway. The local LAN IP address is used as the default
Gateway address for all local nodes wishing to access the Internet. The gateway hides the local IP addresses from the Internet; instead IP datagrams are all sent using the gateway’s WAN or Internet IP address. The gateway keeps track of the local I addresses so that IP traffic from the Internet can be directed to the correct local node. If a LAN includes a gateway that provides a connection to the Internet then the gateway’s LAN IP address must be known to all nodes – in most operating systems this IP address is specified as the default gateway
•Wireless Access Point
Wireless access points (WAPs) or simply access points (APs) are the central nodes on wireless LANs. Access points broadcast to all wireless nodes within the coverage area. On 802.11WLANs the access point does not direct packets to specific nodes or control the order in which nodes can transmit; rather they simply repeat all packets received. Conceptually an access point performs much like a hub on a wired LAN.A significant issue with WLANs is security – any user within the coverage range can potentially access the network. To counteract this possibility access points include security in the form of WEP (Wired Equivalent Privacy) and WPA (Wi-Fi Protected Access). WEP uses a single shared key encryption system whilst WPA generates new encryption keys at regular intervals. The WEP system can and has been infiltrated so currently WPA is the recommended system. No encryption system can work if it is not turned on. This is a major issue for both home and business WLANs. Furthermore the simplicity of creating a WLAN and the ability to access WLANs from outside make security a significant issue. Hackers need only to connect a wireless access point to an existing Ethernet connection point and they then have complete access without the need to work around complex firewalls and proxy servers.
Modem
The term modem is a shortened form of the terms modulation and demodulation, these are the primary processes performed by all modems. Today most modems are used to connect a computer to a local Internet Service Provider (ISP); the ISP supplying a high-speed ADSL or cable connection to the Internet. Dial-up modems were once the primary device for connecting users to the Internet. Currently dial-up modems are more often used to send faxes from computers over the PSTN – virtually all dial-up modems are able to both send and receive fax transmissions. We discussed modulation in some detail earlier in this chapter. Basically modems modulate digital signals by altering the phase, amplitude and/or frequency of electromagnetic waves. That is, modulation is the process of encoding digital data onto an analog waveform. Demodulation is the reverse of the modulation process. Demodulation decodes analog signals back into their original digital form. Clearly both sender and receiver must agree on the method of modulation used if communication is to be successful. Modems are commonly connected to a computer via a USB port or an Ethernet network connection. These interfaces are considered digital links; they do use electromagnetic waves however the data is represented using different voltages. The electronic circuits within the computer can use these voltage changes directly. In contrast modulated analog waves, such as those transmitted down telephone lines or coaxial cables, are not suitable for direct use by the circuits within the computer. Hence the primary role of modems is to provide an interface between the modulatedanalog waves used for long distance transfer and the digital data suitable for use by computers.
ADSL modems
Asymmetrical digital subscriber lines (ADSL) use existing copper telephone lines to transfer broadband signals. Although these copper wires were originally designed to support voice frequencies from 200 to 3400Hz, they are physically capable of supporting a much wider range of frequencies. It is the various switching and filtering hardware devices within the standard telephone network that prevent the transfer of frequencies above about 3400Hz. To solve this problem requires dedicated hardware to be installed where each copper line enters the local telephone exchange. ADSL signal strength deteriorates as distances increase; the signal cannot be maintained at all for distances greater than about 5400 metres. Voice lines much greater than 5400 metres are possible using amplifiers. Unfortunately these amplifiers boost only the lower frequencies required for voice, hence ADSL is not currently available in many remote rural areas. Even when distances are short and the copper runs directly into the exchange problems can occur as a consequence of interference. In general phone lines within a building and out to the street are not shielded against interference, this interference is rarely significant enough that a connection cannot be established; however it often reduces the speed of such connections.
So how does ADSL transfer data between an ADSL modem and the local telephone exchange? Using a modulation standard known as Discrete Multitude (DMT). DMT operates using frequencies from about 8kHz to around 1.5MHz.This bandwidth is split into some 247individual 4kHz wide channels . Each channel is modulated using QAM. DMT’s task is to specify the channels that are used for actual data transfer. If interference is present on a particular 4 kHz channel then DMT will shut down that channel and assign a new channel. This channel switching occurs in real time and is completely transparent to the user. In a sense ADSL is like having 247 dial-up modems all working together, each modem using QAM and DMT ensuring they all work together efficiently. The ADSL modem and the DSL hardware at the telephone exchange communicate to agree on the channels currently being used. At the local telephone exchange all the copper wires from the neighbourhood are connected to a splitter (see
Multiplexor
Simply refers tithe DSLAM’s task of combining multiple signals from customers onto single line and extracting individual customer signals from this single line. In most ADSL systems the lower bandwidth ADSL channels are used for upstream data (from modem to exchange) and higher frequency channels are used for downstream data (exchange to modem). Some channels are able to transfer data in both directions. ADSL is one example of a DSL technology, the A stands for asymmetrical, meaning transmitting and receiving occur at different speeds.
Cable modems
Cable modems connect to the Internet via coaxial cables; usually the same cable that transmits cable TV stations.
Fig 3.92
Describes how the bandwidth within the cable is split into channels. A single 6MHz bandwidth channel is used for downstream data – 6MHz is the width of a single cable TV station. This 6MHz wide channel is assigned within the range 88 to 860megahertz. A narrower bandwidth channel is used for upstream; commonly 1.6MHzwide however various other bandwidths are supported ranging from 200 kHz to3.2MHz. The upstream channel is assigned within the range 5 to 42 megahertz. The particular frequencies used for both channels are determined by the cable Internet provider and cannot be altered by individual users. The bandwidth used in a cable system is significantly larger than that used for ADSL.Therefore, one would assume the rate of data transfer would be much larger. In reality cable connections achieve speeds similar to ADSL connections; why is this? Cable connections are shared amongst multiple users. Single 6MHz downstream channels likely to be shared by hundreds of users. In a sense all the cable modems sharing a particular channel form a local area network. Every cable modem within the network receives all messages; they just ignore messages addressed to other modems. Consequently when only a few users are downloading then higher speeds are possible than when many users are downloading. Clearly the same situation occurs when uploading. This is why cable Internet companies include statements within their conditions stating that speeds quoted are not guaranteed
Router
Routers specialise in directing messages over the most efficient path to their destination. Today the large majority of routers operate at the network layer of this model using the IP protocol. Therefore routing decisions are based on each datagram’s destination IP address. Routers usually include the functionality of gateway. They are able to communicate with networks that use different protocols and even completely different methods and media for communication. Many routers also include variety of different security features. They are able to block messages based on the sender’s Padres, block access to specific web sites and even restrict communication to certain high-level protocols. Home or small business routers connect a singleLAN to the Internet. For these systems the decision is relatively simple – either the I datagram is addressed to a local node or it is not. Local datagrams are left alone whilst all others are sent out to the Internet. The routing table maintained by these routers is relatively small and rarely changes. Home and small business routers are commonly integrated devices that commonly include a router, an Ethernet switch and also wireless access point – these integrated devices are what the general public callrouters.Routers out on the larger Internet connects too many other routers. For these routers deciding on the best path for each IP datagram is considerably more complex. Such routers communicate with other adjoining routers to continually update their internal routing table. The routing table is examined to determine the most efficient route for each IP datagram. However, should any connections within the most efficient path fail then routers automatically direct the message over an alternate path? On larger wide area networks, and in particular the Internet, thousands of routers work together to pass messages to their final destination.
SERVERS
Servers provide specific processing services to other nodes (clients). We discussed the general operation of client-server architectures earlier in this chapter. In this section we briefly consider some of the more common services performed by servers. Note that this section is included under the general heading of “Network Hardware”; servers are often distinct computers designed with hardware suited to the services they provide, however what makes them servers is actually the installed software. On large networks dedicated servers are common whilst on smaller networks a server may well perform many tasks including the execution of end-user applications. Most servers run a network operating system (NOS) to manage user access to the services the server provides. We discuss features of network operating systems in the next section. Most network operating systems include file server and print server functionality as these are the core services that require user authentication and user access rights. There are numerous different services that servers provide. Examples of servers include file servers, print servers, database servers, mail servers, web servers and proxy servers. In this section we restrict our discussion to a brief overview of each of these services.
File Servers
A file server manages storage and retrieval of files and also application software in response to client requests. In hardware terms dedicated file servers do not require extremely fast processors, their main requirement being large amounts of fast secondary storage and a sufficiently fast connection to the network. Commonly file servers include multiple hard disks connected together into an array – RAID (Redundant Array of Independent Disks). Users are often unaware that multiple disks are being used. RAID uses different combinations of striping and mirroring to both improve data access speeds and also to improve the fault tolerance of the system. Striping stores single files across a number of physical disks and mirroring stores the same data on more than one disk. On larger RAID systems it is possible to replace faulty drives without halting the system – this is
Known as hot swapping. To further improve fault tolerance many file servers include various other redundant components including extra power supplies, cooling fans and in some cases the complete server is replicated. File servers must be able to process multiple file access requests from many users. Consequently the network connection to a file server often operates at a higher speed than for other workstation nodes. For each client request the file server, in combination with the NOS, checks the user’s access rights or permissions before retrieving the file. The file server in combination with the NOS ensures the file is retrieved and transmitted according to the user’s assigned access rights
Web Servers
We discussed the operation of web servers when discussing the HTTP protocol earlier in this chapter. Essentially a web server provides services to web browsers – they retrieve web pages and transmit them back to the requesting client web browser. Web servers must also include services that allow web pages to be uploaded, edited and deleted. Such services require users to first be authenticated by the web server. Many web servers, particularly those operated by ISPs, host many different web sites. These servers require high speed links to the Internet together with fast access to the files they host