Static and public IP addresses for regular computers

Question

I want to do a messaging application on the browser using WebRTC, but I want to get rid of every third party like STUN and TURN servers (I also want to get rid of signaling servers but first things first). I want the users to keep their contacts in the browser localStorage in a key-value way: name of the person => IP address.

I don't really care how the users find their own public IP address (they can do an ipconfig as there is no web API to retrieve it) nor how they distribute it (they can use use a centralized service like Messenger to give their public IP address to their friends, or they can use a QR code on their business card).

But the main issue I have is that I want these public IP addresses to be static, because I don't want to notify all my friends to update their contact file every time my gateway changes my public IP address.

In IPv4, there are too few available addresses (only 4 billions), so the public static addresses are all reserved to website and residential gateways. Whenever I want to access the Internet, my gateway opens a specific port for my computer. For example, if my gateway has the public address 1.2.3.4, the "public address" of my computer would be temporarily 1.2.3.4:3000. This process is called NAT. To find one's "public address", one must send a request to a STUN server which would respond with what IPv4 address and what port it sees. But the gateway closes the connection of the port at some point, so that's not a public static address like I want.

But in IPv6 it's different, the number of possible addresses is ridiculously high (2^128), so we could theorically give one static public address to each computer in the world. NAT would basically be useless (I'm not talking about firewalls here). But again, there is a problem, in IPv6 you have one address that is static but not public, and you have one or more addresses that are public but not static. So nothing has really changed from IPv4, and it still doesn't solve my problem.

I have 2 questions:

• How to have a public and static address for regular computers? I don't want my users to meddle with their router/ISP or install anything on their computer.
• Why don't we all already have static and public IPv6 addresses? Is there a design choice behind this?

Thank you for your help.

Answer 1

Your question is not really a programming question, so I'm assuming you need to know this because you are developing software that uses IPv6. Looking at it from a application point of view:

1. You don't need TURN to discover your own public address. Without NAT your address is not changed by the network and your local address is your public address. You just need to let your software ask the OS what the currently configured addresses are.

2. With IPv6 it's usually not the gateway giving out addresses to connected devices. The gateway only communicates the network prefix (a /64) to the network, and all devices choose their own addresses (usually multiple addresses per device).

DHCPv6 does exist, but it is usually not used to configure IPv6 addresses on devices. It may be used in stateless mode to configure domain names, DNS resolvers etc, but that is not related to assigning addresses.

So the device chooses its own addresses within the prefix communicated by the gateway. How does it choose?

Originally devices used their interface MAC address for the suffix (using the modified EUI-64 algorithm). The problem with this was that if you had a lap for example, online services that you connect to could track your device. Because no matter what network you connect to, the last 64 bits of your address would contain your unique MAC address. That way it was possible to track your device from work to home to your favourite coffee place etc.

These days the OS uses privacy extensions to make your IPv6 address untraceable. Often there is a stable address that is linked to the network prefix, so that every time you connect to the same network you get the same address. That address is useful for peer-to-peer communication.

But they went further than that. There is often also a set of privacy addresses that (semi)randomly change over time. Your device might generate a new address every day, or maybe even every hour. That makes it even harden for online services to determine whether there is a single device that has changed it's privacy address, or whether there are multiple devices. Because this address changes over time it's good for short-lived outbound connections, but not so useful to accept inbound connections on. For that use the stable address I mentioned in the previous paragraph.

Probably the most difficult bit is how to ask the OS about the available addresses. Looking at my own macOS box I see:

en2: flags=8863<UP,BROADCAST,SMART,RUNNING,SIMPLEX,MULTICAST> mtu 1500
    ether c8:e0:eb:5c:af:61
    inet6 fe80::1074:8568:e447:d9e3%en2 prefixlen 64 secured scopeid 0x12
    inet6 2a02:xxxx:xxxx:xxxx:3e:873f:837:1417 prefixlen 64 autoconf secured
    inet6 2a02:xxxx:xxxx:xxxx:b19c:71c5:1de8:8fde prefixlen 64 autoconf temporary

You can see both the long-term secured address and the short-term temporary address.

Without knowing which OS you are developing for and in which programming language I can't help you further. Please adjust your question to include that information.