[Guide Home]

The Web Hikers guide to
Outpost Firewall

 

Home > FAQ > The Internet > How the Internet and the WWW work > Structure

Presentation

[Green colours with tiny sized fonts] [Green colours with medium sized fonts] [Green colours with large sized fonts]

[Orange colours with tiny sized fonts] [Orange colours with medium sized fonts] [Orange colours with large sized fonts]

[Blue colours with tiny sized fonts] [Blue colours with medium sized fonts] [Blue colours with large sized fonts]

[Grey colours with tiny sized fonts] [Grey colours with medium sized fonts] [Grey colours with large sized fonts]

Update Alert
Popup Glossary

About Presentation

Structure

Bottom

Many people believe the web to be similar to a spider web, but this is in fact not so. A recent study of 1.5 billion web pages and their relationships has found the web to be more like a bow tie, while the structure of the hardware is more like an hourglass (Laudon, K. C. and Traver C. G. (2002). E-commerce: business, technology, society. USA: Addison Wesley).

The WWW today

In 2001 there were approx. 400 million users of the Internet world-wide (up from 100 million at the end of 1997) and projections suggest that by 2003 the number of users will be 800 million (ibid). How does the Internet manage to keep going with so many users? And how does it adapt to such increases of growth?

First, the client/server relationship is very adaptable, the size of the net can be increased and is increased by having new servers and clients connect to it. Second, the architecture behind the web is made of layers, which means that if part of it needs changing then the change can occur without effecting the other layers.

Layer 1
Network Technology Substrate, this is the cables/switches, the hardware.

Layer 2
These are the protocols, like UDP and TCP, they occupy the middle of the hourglass.

Layer 3
This the middleware layer, this layer allows the Network and application layers to connect to each other. It covers things such as security and nameservers and so on.

Layer 4
Applications layer, these are things like Outlook Express and Internet Explorer, as well as web servers.

Users of applications are rarely aware of the middleware layer, it operates in the background, like when you request a web page and your browser queries a DNS server for the pages IP address. Because of the layered structure their can be major changes carried out in the network layer without the application layer being affected.

Glossary Bottom Top

Internet Architecture

The layout of the hardware is like a pyramid, at the top are the backbones.

Backbones
These are high bandwidth fibre optic cables and satelites going across the world which are formed by NSPs (Network Service Providers) such as BT, AT&T, Cable & Wireless etc., etc.. The US backbones are entirely fibre optic and have a bandwidth ranging from 155Mbps to 2.5 Gbps (Gigabits (billions of bits) per second). The backbones have their own built in redundancy, multiple connections to paths in the network. Backbones are connected to NAPs (Network Access Points) or MAEs (Metropolitan Area Exchanges).

Network Access Points
These are hubs where backbones connect to each other as well as with regional and local area networks. They are owned by private telecoms companies and have a bandwidth at over 100Mbps. The telecoms companies lease access to ISPs (Internet Service Providers - what you connect to in order to access the Internet), private companies and governments.

Campus Networks
After the NAPs come campus networks. There are over 1 million campus networks connected to the Internet today (Computer Industry Almanac Inc., 2001), they are large companies like Microsoft and Universities that are big enough to be able to lease access directly from a Regional Service Provider.

Internet Service Providers
These provide the lowest level of access and service small businesses, home users and some institutions, they are the retail providers. They have connections to the net at up to 45Mbps and offer two types of access for their customers, Narrowband and Broadband.

Narrowband are dial-in modems which operate at up to 56Kbps (Kilobits (thousands of bits) per second) although 30Kbps is average as line noise causes packet loss (that's what all that noise is when you dial in, data echoing and rebounding). The 56Kbps limit is due to America, their phone lines can't carry more the 56Kbps but British and European lines can take up to 60kbps but to keep compatibility the speed is capped in line with America.

Broadband is Cable, DSL, T1 and T3 connections. These are 'always on', no dialling is required. As far as Internet technology is concerned Broadband is generally anything that offers 100Kbps or more or put another way, a comms technology that allows smooth delivery of streamed media.

  • DSL operates at between 150Kbps to 1Mbps (Megabits (millions of bits) per second).
  • Cable operates at between 350Kbps and 1Mbps.
  • T1 and T3 are international telephone standards with guaranteed delivery at 1.54Mbps (T1) and 43Mbps (T3).

To help you visualise what these speeds mean here is how long it takes the different types of connections to download a 10 MegaByte file:

  • Telephone Modem 56K: 25 Minutes
  • DSL lite: 9 Minutes
  • DSL regular: 3.5 Minutes
  • Cable modem: 3.5 Minutes
  • T1: 51 seconds
  • T3: 3 seconds

Glossary Bottom Top

Problems with the Internet

A lot of the Internet architecture is decades old (about 100 years in Internet time) and this is the cause of insufficient bandwidth at all levels from the backbones to the ISPs. This means the service is slow and can't handle video or voice traffic very well.

The way packets are handled is also a problem, and this problem creates a situation known as latency - the time it takes for packets to travel through the network. If all you do is send e-mail then you probably haven't experienced latency, but try playing a game or watching a film over the Internet and you will soon discover what latency is. The Quality of Service offered today is 'best effort', no guarantee is given that data will arrive or when it will arrive, also packets have the same priority no matter what they are sending or who the data is for so an e-mail which has low priority currently has the same priority as a streamed film which should be given a higher priority if the level of service is to be maintained (ibid).

Another problem is Unicast Streaming. Pretend that you have just requested a streamed music video and it gets sent to you in packets. Now imagine 1000 other people also requesting the same file, the video is dispatched over 1000 times, the duplication of packets slows down the network even though they are sending the same information. Compare this to other streamed media like radio or television where the data is only sent once to millions, and in some cases (such as the World Cup) billions of houses. If only the Internet could be that efficient, can you imagine the speed increase!

HTML is another limiting factor (ironic when you consider it was HTML that created the Web in the first place). It is fixed and generic, it's OK for static text and images but not for communicating richer documents like databases graphics or business documents.

What you need to fix the problems of the current Internet is a faster more powerful version - welcome to the world of the Internet2® Project!

Internet2®

Glossary Top

 

Outpost and the Outpost logo are ©Agnitum Software

This is an unofficial guide, the information expressed here may differ from Agnitum's. There is a support forum (no longer run by Agnitum, but by users) if you need more help this is a good place to start. Where information here conflicts with what Agnitum have told you always go with the information given to you by Agnitum.

 

Guide/site and images ©Stephen Cox