- Physical Network Topologies
- Logical Network Topologies
- Communication Methods
- Network Devices
A network is set of computers linked together for the purpose of communicating and sharing information. The Internet is a global super-network, so is the local area network ( a LAN) at your workplace or your school, as is the wireless hotspot at your local coffee shop, hotel or library, the telehone and cellular systems, and the satellite communications in space.
You know you have a network when you have two or more computers connected together and they are able to communicate. Plugged into the back of each computer is some sort of communications port. Nearly all computers today have one or more serial ports, parallel ports, Ethernet ports, modem ports, firewire ports, USB ports and more. All of these ports can be used in one way or another to connect computers to a network. The most common type of network port is an Ethernet port (the square port with the row of connectors on the bottom). The next most common is a wireless network connection, but that has no physical connector port.
Xerox was the first company to research and develop a network. Once upon a time, Xerox printers were extremely expensive, so companies wanted to share them. Xerox knew their printers were expensive and users were only able to print from one big computer (a mainframe) attached to the printer directly. Xerox decided that they could sell more printers if they could make it possible for anyone to use the printer from any computer. To allow multiple computers to communicate with the printer, some means of sharing a connection to the printer was needed. Xerox put Bob Metcalf and others to work on researching and designing what eventually came to be called ethernet.
Hosts, End Stations and Workstations
When people talk about networks, they often refer to computers that are at the edge of the network as hosts, end stations, workstations, or servers. Its all just the same thing, a computer attached to the network; though the word HOST has the most general meaning and can include anything attached to the network including hubs, bridges, switches, routers, access points, firewalls, workstations, servers, mainframes, printers, scanners, copiers, fax machines and more!
Just about everything electronic that has a processor and which you would use in an office is 'network capable' today and lots of things that aren't currently networked probably will be networked in the future. In many offices the phone system already IS the network (Voice over IP).
LAN, MAN, WAN and er.. IPAN??
A Metropolitan Area Network (MAN) is a network that encompasses a city or town. It is usually multiple point-to-point fiber-optic connections put together by a communications company and leased to their customers, but a small number of big corporations have built a few of these of their own and opened them to the local companies with which they do business. The automotive, travel and insurance industries are just a few examples of who has built a WAN.
A Wide Area Network (WAN) is usually composed of all the links that connect the buildings of a campus together, such as at a University or at a corporate headquarters. WAN connections can often span miles, so you frequently hear peole referring to the 'WAN' connection to an office half way around the world. Usually, what distinguishes a WAN from a LAN is that there are one or more links that span a large distance over serial, T-carrier or ISDN, Frame Relay or ATM links.
So what the heck is an IPAN? An IPAN is an Inter-Planetary Area Network. NASA has built a Deep Space Internet that uses a store-and-forward communications protocol called Disruption Tolerant Networking (DTN). The mechanical rovers Spirit and Opportunity on the planet Mars, have addresses on a NASA network and NASA uses Internet protocols to communicate with the Mars rovers. While the communication with the Spirit rover doesn't actually get transmitted over the Internet, the NASA network does have hosts spanning between the planets Earth and Mars. They also have probes they have sent into the outer solar system
From a very practical and minimalist standpoint, we can start with the most common type of network, an Ethernet based local area network as an example. The network is built with several layers of technologies piled on top of one another.
- Physical communication using electricity over copper wiring
- Physical addresses (Media Access Control or MAC addresses)
- Dynamic Host Control Protocol - DHCP
- Provides a logical network address (an IP address)
- Address Resolution Protocol / Reverse Address Resolution Protocol
- Allows computers to announce a logical address (IP address) they are looking for and to discover which physical address (MAC address) is using that address.
- Internet Protocol
- Logical Addressing
- Transmission Control Protocol or User Datagram Protocol
- Sockets and Sessions
- Domain Name Service (DNS)
- Host name and IP Address resolution
- Other services and protocols (such as HTTP which makes the World Wide Web possible)
All networks have several layers of functions stacked on top of each other. Ethernet is used to provide the means to transmit information encoded in electrical signals across copper wiring between two computers. Internet Protocol networking software running on the computers use the Ethernet network to send data back and forth inside IP packets. The Internet Protocol layer provides the means for the computer to connect to the network, obtain a logical address, to learn the logical addresses of other computers and to communicate with the other computers on the network. Internet Protocol provides the basic network functions.
The hardware used to build the network will usually require that the structure of the network conform to a certain design. The word topology is used to describe what the network looks like when drawn on paper and to a large extent, how it operates.
- Bus Topology
- A bus topology connects all computers together using a single wire, usually a piece of coaxial cable, that passes electricity over a copper core that all devices transmit and receive from. All devices hear all communication over the bus.
- Ring Topology
- A ring topology usually involves connecting one or more computers together using paired physical interfaces. One interface is the clockwise side of the ring, the other connection is the counter-clockwise side of the ring. Devices connected to the ring can transmit and receive, but there is usually some other sort of method for controlling access to the common network hardware. Token Ring uses a ring topology as does CDDI and FDDI. All three of these network technologies use a token-passing scheme in which the computer holding the the token is allowed to transmit.
- Star Topology
- A star topology is the most common network topology in use today. All devices in the network are connected to a single hub or repeater. The connected devices radiate outward from the hub like an asterisk '*' or star.
- Hub and Spoke Topology
- Hub and spoke is another term often used to describe a star topology.
- Point to Point Topology (Daisy Chaining)
- A point-to-point topology is most often a communications connection between two devices over a single hardware connection that is not shared by any other devices. There will be exactly two and only two devices on the connection. Networks using point-to-point topologies can be daisy-chained together to form an end-to-end communications path.
- Point to Multipoint
- A single connection point on the network has network segments that run to several other points.
A peer-to-peer network is composed of two or more self-sufficient computers. Each computer handles all functions, logging in, storage, providing a user interface etc. The computers on a peer-to-peer network can communicate, but do not need the resources or services available from the other computers on the network. Peer-to-peer is the opposite of the client-server logical network model.
A Microsoft Windows Workgroup is one example of a peer-to-peer network. UNIX servers running as stand-alone systems are also a peer-to-peer network. Logins, services and files are local to the computer. You can only access resources on other peer computers if you have logins on the peer computers.
The simplest client-server network is composed of a server and one or more clients. The server provides a service that the client computer needs. Clients connect to the server across the network in order to access the service. A server can be a piece of software running on a computer, or it can be the computer itself.
One of the simplest examples of client-server is a File Transfer Protocol (FTP) session. File Transfer Protocol (FTP) is a protocol and service that allows your computer to get or put files to a second computer using a network connection. A computer running FTP software opens a session to an FTP server to download or upload a file. The FTP server is providing file storage services over the network. Because it is providing file storage services, it is said to be a 'file server'. A client software application is required to access the FTP service running on the file server.
Most computer networks today control logins on all machines from a centralized logon server. When you sit down to a computer and type in your username and password, your username and password are sent by the computer to the logon server. UNIX servers use NIS, NIS+ or LDAP to provide these login services. Microsoft Windows comptuers use Active Directory and Windows Logon and/or an LDAP client.
Users on a client-server network will usually only need one login to access resources on the network.
Computer networks using distributed services provide those services to client computers, but not from a centralized server. The services are running on more than one computer and some or all of the functions provided by the service are provided by more than one server.
The simplest example of a distributed service is Domain Name Service (DNS) which performs the function of turning human-understandable domain names into numerical (dotted quad) computer addresses called IP addresses. Whenever you browse a web page, your computer uses DNS. Your computer sends a DNS request to your local DNS server. Your local DNS server will then contact a remote server on the Internet called a "DNS Root Server" to begin the lookup process. This DNS Root Server will then direct your local DNS server to the owner of the domain name the website is a part of. Thus, there are at least three DNS servers involved in the process of finding and providing the IP address of the website you intended to browse. Your local DNS server provides the query functions and asks other servers for information. The Root DNS server tells your local DNS server where to find an answer. The DNS server that 'owns' the domain of the website you are trying to browse tells your local DNS server the correct IP address. Your computer stores that IP address in its own local DNS cache. Thus, DNS is a distributed service that runs everywhere, but no one computer can do the job by itself.
Today's network environment is predominently Ethernet technologies. Ethernet is a broadcast protocol that provides the physical layer and data-link layer functions within a network. To connect devices that use ethernet, you need a hub, bridge, switch or router, and which device you use depends on how you will use the network and the computers.
A hub is a device used to create a broadcast domain so that several computers can communicate. Hubs are very inexpensive. Hubs receive frames and broadcast the frame on all ports. Hubs are dying out of existence and the only place you see them today is in extremely small networks, especially in homes. Once there are too many computers connected to a hub, communication begins to break down and a more intelligent solution is required.
A bridge is a smarter version of a hub, and performs the same function. A learning bridge can figure out where a given computer is located and transmit frames only on the port connected to that device.
A switch is a specialized type of learning bridge that can learn which devices are connected to which ports and can forward frames only to those ports that are supposed to hear the transmission. Bridges maintain a forwarding table. Switches are the most common way in which ethernet networks are wired together.
Routers break up broadcast domains and segment networks, which allow network administrators to control broadcasts and control access to various network resources. Routers provide a means to allow computers to share a logical network that is separate from the physical switched network. Routers are used to connect several ethernet networks together to make a larger network. Routers can further extend the network by connecting local networks to the global network called the "Internet".