Home / server / Questions / 39271
Accepted
PostulatedAxiom
Asked: 2009-07-12 02:38:12 +0800 CST

How does the internet really work?

  • 772

Sounds like a dumb question but I bet a lot of people don't know either. I understand servers, clients, modems, routers, ISP's, ect; but what I don't understand is what makes up the backbone structure of the internet. I have never seen any clear UML diagram or description of the backbone of the internet. I have heard things about 7 main servers (don't quote me on that), but who owns each server, when were they built, how old are they, how do they interact? It seems surprisingly hard to find information on this.

All my google searching provides seemingly vague and outdated information.

Edit: Sorry if you found this question vague, not only did I write it late last night, but I have a vague understanding of how the backbone of the net works thus making my question vague.

internet net

6 Answers

  1. Best Answer

    I'll take a stab at this:

    First thing first, no one owns or controls the internet. Right now there is defacto control provided through the DNS servers, which are what change "www.google.com" into "IP address 123.456.789.000". These DNS "root" servers control the domain name infrastructure that provides the web as many people know it.

    However the internet is actually a network of networks controlled by people (hence inter - network). If you imagine that you have a network of computers controlled by a cable provider, by a telephone provider, network them to a government network, network them to a link to europe, another to hawaii, another to asia, another to Australia, you can see how the internet starts to take shape. Essentially companies and in some cases countries will pay to have their internet connected to a link in a network to america. Once these links were established the internet really started to take shape.

    From a hardware point of view, the internet is built upon IP (not TCP/IP). IP was a system that provides for networking using a shared address space (the familiar www.xxx.yyy.zzz) addressing system with a notion of a "gateway" which is if I don't know the person who owns this packet I'll forward it to someone who does. Essentially there was a routing network created which defines which routers control certain IP ranges. That way if I can digress, for your computer to send a packet in america to a computer in australia, the following would happen.

    1. You send the packet via your modem to your ISP.
    2. Your ISP uses it's rules to determine it doesn't own the IP of the packet and forwards it to it's backbone or a Tier 1 provider.
    3. The backbone will determine that this packet is for australia and sends it to a machine connected to a fibre link or possibly over a satellite etc.
    4. This process happens in reverse from the backbone, to the ISP, to the local ISP connection, to the modem in the house, to the computer in the house.

    Now when you realise that the routing rules have redundancy (ie. you have more than one route to send a packet to australia for instance, picking a different cable, or using a sattelite) you can start to understand how the internet can survive when computers or routers shut down or fail, which is a key part of the infrastructure.

    So if you have a network that can get packets to anyone who is connected, and anyone can connect by agreement to tier 1 connections, you can combine the capability to talk to any computer on the network with a protocol to send information, such as HTTP, FTP, SSL etc. you end up with the internet as it exists.

    A final word: If you managed to soak all of that in, you can see now that the argument between "well everyone should be able to watch youtube and make VOIP calls without it EVER being throttled" doesn't mesh so well with the fact that the people who are providing the internet have to share it with networks which they don't control. I'm speaking about net neutrality of course.

    • 24

  2. It's really a series of tubes.

    • 11

  3. I guess you're talking about root servers. These one are used for DNS, without them you would not be able to connect to computers using a name.
    Your question is a bit complex to answer because the term Internet is a large things. If we only consider IP level, Internet is just a lot of routers interconnected together.

    • 4

  4. The backbone of the internet is composed of loooong strands of optical fiber, which interconnect different datacenters (IXP - Internet eXchange Points), where different ISPs keep their routers, make peerings (interconnections).

    So, "who owns the internet?" - well - ISPs, who buy the routers, switches (usually from Cisco), ask other firms to make new connections (via optical fiber) to other PoPs (Points of Presence) or other ISPs (peerings) and implement new protocols (IPv6, multicast transmissions).

    Some interesting links:

    http://en.wikipedia.org/wiki/Internet_exchange_point

    http://en.wikipedia.org/wiki/List_of_Internet_exchange_points_by_size

    Later edit: You may also want to read about the BGP - the main routing protocol for the "core" of the internet - http://en.wikipedia.org/wiki/BGP

    • 4

  5. in a word cisco - lol.

    in reality it works simply on lot of pipes connection oodles of computers all mostly passing on anonymous bits of data till it get's (usually) to the right place. Some are big computers some are small some pipes are wide and fast others are slow. But in essence its all just one logical forwarding system. A bit like a really fast postal system.

    • -1

  6. Pigeons. We'd be lost without the little guys running the packets around. See RFC 1149 and RFC 2549 for more information.

    • -2