| Wikipedia | Introduction to Networking |
Distinction: LAN and WAN (Local Area Network and Wide Area Network). LAN uses protocol such as Ethernet to link computers, WAN uses telecommunications. LAN communications generally much quicker, 100 Mbits/sec or 1GBits/sec as opposed to 128Kbits or 1 or 2 Mbits for a leased line. WAN can use normal line with modem, needs dial-over public telephone network, and is not 'always on'; alternatively can use a leased line, which is dedicated, faster and 'always on'.
A MAN (Metropolitan Area Network) is a hybrid network containing elements of a LAN and a WAN. See the example of London. There are further examples on this site.
A PAN (Personal Area Network) covers immediate workspace, for example a desktop using a localised network such as Bluetooth or a house using Wi-fi.
Star - computer or hub or switch at centre - the actual shape may not be very star-like, hub in a cupboard, computers in rows. Individual cables for computers so failure of one will not bring down the rest. Can cope well with heavy loads.
Bus - single cables from server with computers linked by a 'T' piece. Cable failure brings down all computers. Not good at heavy loads.
Ring - Computers arranged in ring, links from one to the next. Cable failure brings down the network. No collisions. Very fast.
Peer-to-peer
Typical LAN now has bus sections between hubs or switches (the 'backbone) and local star sections.
Network applications: email, EFT (electronic funds transfer), video conferencing, Internet, distributed systems.
Note details of an nternet URL: http:// (protocol); www.address.co.uk (web server or domain name); /main/subject/index.html (folders and files).
A set of connections between geographically remote local area networks. USE: PSTN; dedicated leased lines (high speed, high bandwidth); high speed fibre optic cable; microwave; satellite; radio waves; the internet.
Mobile phone network is a WAN: uses circuit switching for voice calls (complete path through network for a connection) and packets for text messaging (routed). Text messages are put into the packets that mobile phones continually send so base stations can locate them. These packets have 160 bytes spare capacity.
Packet switching on the early internet
Two LANs connected by WAN = internetworked = an internet. Present day Internet grew from wide area connections between university campuses of Stanford, MIT, Carnegie Mellon and others.
Internet Structure Network Layers:TCP/IP & OSI IP Addressing
IP Addresses Explained DNS Explained DNS Explained CompTIA Network SOPA & PIPA
NIC converts data from form stored on computer to that transmitted or received on a cable.
Data in a buffer; checksum calculated; address information added for source and destination (each network card has its own unique MAC address - 6 bytes, 'media access control'). The block of data and information is known as a frame.
Computers and other devices transmit their data along the length of the wire to which they are connected and can be received by any other attached device. With more than one computer transmitting data collisions can occur between pulses from different devices. When such a collision occurs the bus becomes unusable; to reduce the effect of this the transmission duration is limited to one frame of pulses. To reduce the effect of this transmission duration is limited to one frame of pulses.
Connected computers must follow a protocol when transmitting to handle collisions: CSMA/CD (Carrier Sense Multiple Access/Collision Detection). Rules:
Segmentation is used on a bus network to improve efficiency - split a large network into smaller ones to reduce collisions. The smaller networks are linked by routers or bridges. A bridge holds the MAC addresses of network cards on the segments joined by the bridge. A router holds a table of IP addresses, which are more flexible than MAC addresses and can be assigned to user accounts as well as machines.
Broadcasts by computers on a bus network can be read by all other computers on the same segment so computers are not secure against eavesdropping.
Even though the physical topology is a star with computers connected to a central hub or switch, the LAN still behaves as a bus. The central switch queues frames until each one can be placed onto the backbone. The switch ensures that collisions do not occur by buffering.
To ensure bi-directional communication the wire connection between computers and switch include two independent pairs of wires. A workstation transmits a packet of data to the line card (in the switch) along the input pair. The packet is stored in the input buffer of the line card. The switch reads the destination of the packet and routes the packet along the backbone connected to the destination.
Links between the central switch and each computer are independent so computers cannot eavesdrop. Other computers are unaffected if one link fails. In a coaxial cable bus network a broken link brings down the whole segment. If the central computer or switch fails then the network fails. If a domain server linked to the central computer or switch fails or if its link fails then that service will fail too.
This is when a number of terminals or thin clients are connected to a central server, which is an application server, a file server and a domain controller. Domain controller validates users when they log on; file server stores user files; application server runs applications such as office applications. Each thin client has a copy of software running on the server. The thin clients transmit key strokes to the application server.
A wireless network is broadcast so is potentially less less secure than a wired network unless encryption is used. Wireless networking allows connections to a wired network in difficult areas e.g. a listed building or a library - removes the need for cabling. WAPs are designed to use radio frequncies. Mobile phone networks are wireless. The standard for wireless connection is Wi-Fi - the trademark for popular wireless technology e.g. 802.11a, 802.11b, 802.11g, 802.11n.
Wireless networks are more vulnerable to breaches of security (hacking). Weaknesses in WEP (Wired Equivalent Privacy) were discovered in 2001 and WPA (Wi-Fi Protected Access) is now preferred.
Bluetooth is a wireless protocol for short distances e.g. on a desktop: a PAN or personal area network. Latest Bluetooth broadcasts up to 150m so the restriction is no longer true. Used to form a connection between specific devices.
No dedicated servers. All computers function as equals, as servers and clients. Each computer/user can determine what resources are shared with others e.g. files, printer. Can assign passwords to resources such as files but there is no centralised security policy.
Peer to peer LAN is a good choice when there are fewer than 10 users, the users are close (e.g. in a house or office), security is not an issue and future growth is limited.
P2P can be used across a WAN such as the Internet e.g. file sharing - Gnutella, BitTorrent or Knapster where temporary connnection between computers are set up. BitTorrent protocol: large file split into parts, each part is sent to a different computer and the parts are sent between these computers in peer to peer fashion. The large file is sent only once so demand for bandwidth is spread over a longer time. Skype is also a P2P protocol.
P2P network not adequate when more than 10 users. Server functions only as a server. Service requests from network clients. Ensure security of files & directories. Domain controller holds user account details and authenticates users during login. File server stores users' work. Web servers distribute web files e.g. http, ftp.
'Web 2.0' is not an exact or scientific term. In Web 2.0 the internet becomes a means for delivering software as a service.
Google: an application that performs a search. Payment is through advertising or pay to have Google list your business on the first search page. No scheduled software releases, just continuous improvement. No licensing, just usage. No porting to different platforms, just a massively scalable collection of commodity PCs running open source OS with applications and utilities written by Google that are never seen outside the company. Google's data centres are not accessible to the public. Software is never distributed to individuals to run on their own PCs. The service is not a server but it is a service delivered by a massive collection of Internet servers. It is not a browser and it does not host content.
iTunes: an application that runs on a handheld device and reaches out to a massive Web backend.
Facebook: an application that performs social networking. Payment is through advertising, which uses the personal information to target specific groups e.g. engaged women looking for wedding dresses.
Web services are self-contained modular applications that can be acquired over a network. Everything is a service. An API is published for use by other services that encapsulates implementation details.
SaaS: Software as a Service. Eliminates the need to install and run the application locally, so no maintenance. Cost is an ongoing charge rather than a one-off payment. Can prevent copying and protection of IP. Service application may be on vendor's own servers or on a third party's servers. May remove the need for servers for the users.
Traditional systems architectures may have brittle links between components: a company wide system made up of modules may require a lot of work to maintain system integrity. Web systems are also tightly coupled. Redesign of any subsystem may cause the whole to break as the flow of data and control is broken. The tendency to break will be increased by increases in scale, demand, volume and rate of business change. This will engender constant maintenance and replacement of systems as they struggle to cope with rapid change. There may be: unresponsive web sites; lack of speed to market with new products and services; inability to shift to new business opportunities; cometitive threats. Tight coupling of subsystems makes it impossible to deal with change and the dynamics of Web business.
These problems have been solved by changing to web services architecture, a more flexible architecture with greater capacity for change.
Growing Trend Towards Web Services
Best Practices for SOA Governance 1
Web services systems promote decoupling of components and dynamic binding instead. All components in a services system are services: they encapsulate behaviour and publish a messaging API to other collaborating components. Applications under SOA are built by discovering and orchestrating services available across the network or by just-in-time integration of applications.
Web services are small programs that use the messaging API to send messages between larger programs. They can be known as 'middleware' as they link other programs together, thus allowing system integration on a mobile platform. Middleware is defined as software that can provide services to applications in addition to those provided by the operating system. It is particularly relevant to distributed applications and provides links between them. Middleware was used to link older applications. "Services that can be regarded as middleware include enterprise application integration, data integration and message oriented middleware(MOM)." (Wikipedia)
Group of interrelated web development techniques used to create interactive web applications which can retrieve data asynchronously and refresh part of a page rather than having to reload the whole web page every time an update is required. HTTP is a simple protocol that loads an entire page with GET. Ajax allows a page to request only the part that needs to be updated, especially data such as stock prices.
Router receives packets or datagrams from one host or router and uses the destination IP address that the packets contain to send them to another host or router. A router has two IP addresses, one for identification by datagrams coming from other routers and the other for identification by datagrams coming from hosts on the LAN.
Routers use the network part of the destination IP address - the route chosen is determined by this destination IP address. Router is usually connected to a network of other routers. Datagram may pass through several routers. Each router keeps a table of routes to various destinations - like you know which way to turn at a junction on a familiar journey. Datagrams go from router to router as you go from junction to junction. Routers for a given country are arranged in a hierarchy: organisation, local, regional, national (London Internet Exchange: LINX). Each router keeps a table of other routers, computers and networks that it is connected to and enough information to route a packet onto a path that will lead to the required destination.
| The Internet's First Router | LINX Adds Network Capacity | Europe's Largest Data Centre |
The IP address range 202.0.0.0 to 203.255.255.255 has been allocated to the Asia-Pacific region. The routers from an organisation in the UK will use the IP address and their internal tables to send packets to Asian addresses: the local and regional routers will know that the address is not a UK one so they will send the packets to the national router, which will then use the address to send the packets on the correct route to Asia.
An IP address defines a host on the Internet. The Internet Assigned Numbers Authority (IANA) coordinates global IP addresses and routing. IPv4 has 2^32 addresses, expressed as 32 bits in octet form; IPV6 has 2^128 addresses, expressed in hexadecimal. IP addresses are assigned by ISPs, which get them from a Local Internet Registry (LIR), Regional Internet Registry (RIR) or a National Internet Registry (NIR).
Public addresses are routable. Non-routable addresses are used for hosts on LANs and hidden behind routers. Hosts on LANs do not need their own public IP address as they use the LAN's router IP addresses. Hosts on a LAN use network address translation (NAT) to route traffic to hosts with private addresses - like flats in a block with the same street address: flat 1, flat 2, etc. These IP addresses are used for private, non-routable addresses:
10.0.0.0 to 10.255.255.255
172.16.0.0 to 172.31.255.255
192.168.0.0 to 192.255.255.255
| IP Address Class A, B and C | TCP IPv4 Subnetting | Subnetting Made Easy |
| Hub, Switch and Router | IPv6 | The day the routers died |
| OSI Model |
At each stage in the connection the IP addresses stay the same but the source and destination MAC addresses change so that the packet can travel to the next link.
A packet will go from host to LAN router NIC, to external router NIC, to the next router's NIC, to router's LAN card to destination host. At each hop the IP address of the source and destination will stay the same but the MAC address of the next target device will change so the packet can be sent by the router to the next stage of the journey.
A gateway is used to connect networks that use different protocols. They convert frames from on format to another.
Each host has to have these items set in its networking or TCP/IP settings: IP address, subnet mask, gateway or router address & DNS server addresses.
The subnet mask defines the size of the network and tells a host what network it is connected to and so which hosts it can communicate with directly. If a host destination address is out of the immediate network then the packet will be sent to the network's switch or router. The subnet mask for most small LANs is 255.255.255.0.
The gateway or router address is the address of the machine that connects to the next hop on the internet, beyond the local network. If the router knows where to send the packet it forwards it, otherwise it sends it to the next router.
A domain name server (DNS) keeps track of the correspondence between domain names and IP addresses.
| History of the Internet | DNS Explained | DNS Explained |
| Internet Structure | IP Addressing | RSA Animate:Internet in Society |
| Evgeny Morozov Lecture | Nicholas Carr:The Shallows | SOPA & PIPA |
See more links
