This repository contains a specification for generating unique and reproducible IP addresses. The spec supports both version 4 and version 6 addresses.
Any IP address generated for any given thing with a unique name in a particular subnet has a very low probability of colliding with another IP address generated with a different name in the same subnet. This probability will obviously depend on how many IP addresses are available in that particular subnet. As such we recommend the following:-
- Prefer IPv6 over IPv4 wherever possible
- Use a network prefix that gives you the largest address space possible. That is, the smaller the network prefix the better. For IPv6 we recommend that you use a
/80block or lower (this includes the popular/64and/48blocks).
Any given thing with a particular name in a particular subnet will always have the same IP address generated for it. Because of this property, any risk of generating a duplicate IP address can be mitigated in advance via simulation.
- Manually assigning IP addresses is frustrating and error prone, especially as the network grows bigger
- Randomly generated IP addresses have a higher probability of colliding and no way to determine these collisions upfront
- Centrally managing IP addresses, e.g. using service discovery tools or storing in a database, does not only introduce more complexity in deployment, it also introduces overhead (both network and storage) and worse still that central system is a potential point of failure
- This spec introduces a standard way for both assigning IP addresses in your network and utilising them in a reliable manner. See our example use case to have a better understanding of how this can be helpful.
This is an open-source, community-driven project, developed under the Apache 2.0 license.