Networking 101: Understanding IP Addresses - Page 2

 By Charlie Schluting
Page 2 of 2   |  Back to Page 1
Print Article

There are also some special cases of IP addresses, including private and multicast addresses. Addresses in the range - are reserved for multicast. Everything below that range is fair game on the Internet, excluding addresses reserved by RFC 1918 and a few other special-purpose assignments. These "1918 addresses" are private addresses, meaning Internet routers will not route them. The ranges include:

  • -
  • -
  • -

    These IP addresses can be assigned locally to as many computers as you want, but before those computers access the Internet, the addresses must be translated to a globally routable address. This is commonly done via Network Address Translation ( NAT ) (define) . The 1918 addresses aren't the only reserved spaces, but they are defined to be "site local." Multicast also has a reserved range of addresses that aren't designed to escape onto the Internet: - are multicast "link-local" addresses.

    To give the necessary background for the next issue of Networking 101, we need to make sure everyone understands the concept of a local subnet. Once we have assigned a valid IP address to a computer, it will be able to speak to the local network, assuming the subnet mask is configured properly. The subnet mask tells the operating system which IP addresses are on the local subnet and which are not. If an IP we wish to talk to is located on the local subnet, then the operating system can speak directly to it without using the router. In other words, it can ARP for the machine, and just start talking. IP address and subnet mask configuration is fairly straightforward for general /24 networks. The standard mask means that the first three octets are the network address, and the last part is reserved for hosts. For example, a computer assigned the IP of and a mask of (a /24, or 24-bits if you write it out in binary) can talk to anyone inside the range

    Be sure to digest everything here, because next we'll get to the meat of subnetting with CIDR.

    In a Nutshell

    • IP addresses are just 32-bit numbers. Subnet masks are just a "cover" that can be arbitrarily slid up and down the IP address's bits to create larger or smaller networks.
    • The "network" portion of an IP address tells the host how large its local subnet is, which in turn tells it who can be spoken to directly.
    • Unicast packets go to one computer, broadcast packets go to many.

    When he's not writing for Enterprise Networking Planet or riding his motorcycle, Charlie Schluting works as the VP of Strategic Alliances at the US Division of LINBIT, the creators of DRBD. He also operates OmniTraining.net, and recently finished Network Ninja, a must-read for every network engineer.

  • This article was originally published on Oct 2, 2007
    Get the Latest Scoop with Networking Update Newsletter