Personal tools
You are here: Home Internet Information Getting IP Numbers in New Zealand

Getting IP Numbers in New Zealand

A guide to 'Getting IP Numbers' in New Zealand.

Note: InternetNZ does not issue IP numbers. The following information is provided purely as a public service.

IP Numbers

If you want your computers to communicate with other computers over the Internet, they will need to have Internet addresses, called IP numbers. (IP is short for Internet Protocol.)

Since late 1995, IP numbers have been handed out in New Zealand using a system called "provider-based addressing". Generally speaking, you get your IP numbers from whoever you connect to the Internet through. This applies whether you are an Internet Service Provider or an end user of Internet services.

Getting IP numbers in this way is usually a simple process. But it does mean that if you change Internet providers, you may need to change your IP numbers. In a small organisation, this is a small problem. In a big organisation, it can be a big problem. There are not many alternatives for those who don't want just to use IP numbers allocated by their provider.

Do please bear in mind that over the next few years only a fixed number of IP addresses are available for everyone in the world. This means that anyone giving out numbers is required to make sure that IP numbers are used efficiently. They are likely to ask you to justify the number of addresses you ask for.

For more information on how IP numbers are allocated, see RFC2050

A Drop of CIDR

One of the big jobs which has to be done to make the Internet work is the distribution of routing information. That is, passing around the Internet information about what path to send data along so that it arrives at a given IP number.

Until 1994, this was done throughout the Internet using "class-based routing". This required that IP numbers be allocated to organisations using a system of "classes".

An IP number is made up of four bytes, usually written like this: 

A "class A" number had only the first byte of the address fixed, leaving the other three bytes to be used by the holder however they liked. An organisation using a "class A" address had enough IP numbers for nearly 17 million devices.

A "class B" number had the first two bytes fixed. So where a "class A" number could be written as a single number, say 18, a class B number had two, as in 130.217. A "class B" number provided address space for 65 536 devices.

The other class was "class C", where the first three bytes are fixed, as in 130.217. A "class C" address allowed room for at most 256 devices.

The trouble with allocating IP numbers in this way was that it was inefficient, often requiring an organisation to take many more numbers than it really needed - if you had five hundred computers, you needed to take a "class B" address. As use of the Internet increased, it became clear that there soon wouldn't be enough numbers to go round.


The answer was Classless Inter-Domain Routing, or CIDR for short (pronounced "cider"). This allowed routing information to be passed around in a way which did not require IP addresses only to come in three types. So instead of an organisation with 500 computers asking for a "class B" address, it would ask for the equivalent of two (or probably four) "class C" addresses together in a "CIDR block".

Since addresses weren't being issued by class any more, a new way was needed to describe them. When asking for a CIDR block of addresses, you specify how many bits of the address are fixed. For example, what used to be called a "class C" address, with twenty-four bits fixed and only the eight bits at the end varying, is now called a /24. To request two of them together, you'd ask for a /23. For four together, a /22.

There are now many more different sizes of "CIDR block" which can be allocated than there used to be, which carries with it a greater need for applicants to justify the size of allocation they ask for.

You have two other alternatives if you do not wish to use IP numbers allocated by your provider.

Address Translation

You can use address translation. This would mean that inside your organisation you have your own IP addressing scheme. Machines which connect your internal network to the wider Internet then carry out a translation process, so that computers inside your network can talk to those on the outside and vice versa. Computers outside your network would never see your internal numbers. Since only the devices at the boundary of your network need to have IP numbers from your provider, renumbering would not be a major task.

The translation process does add complexity to your communications, though.

If you are a large organisation, you have the option of getting IP numbers direct from the organisation which issues IP numbers for the whole of Asia and the Pacific. This is APNIC (Asia Pacific Network Information Centre), headquartered in Tokyo.

For more information on setting up your own internal IP number scheme, visit the APNIC website. 

APNIC website

IP numbers issued by the APNIC do not need to be changed when you change Internet providers, but APNIC charge for their allocation services.

There is no set minimum allocation APNIC will make. Quite apart from the cost, though, some long-distance carriers may restrict how small a block of addresses they will carry routing information for. This means it may be dangerous to step outside the use of IP numbers from your provider for a block of numbers any smaller than a /19, that is thirty-two Class C's in the old language.

Document Actions