13

We are currently building a networking course with a practical part where students have to configure a small network in a virtual environment.

However, we are not sure what ipv6 prefixes we can use for the configuration. The IPv6 documentation prefix (2001:db8::/32 RFC3849) seem to fit. However, someone brought up that this prefix should only be used in textual examples and never be actually configured anywhere, not even in an isolated test network. (Of course it would work, but could be seen as RFC violation).

The issue here is that the RFC is rather abstract about this and it all depends on how one interprets the term "documentation". I couldn't find any sources about the usage in experimental test/tutorial networks or the specification of "documentation" on the internet.

Can you help me out?

11

If your test environment is separated from your local network you hypothetically could use any IPv6 addresses which are not reserved for special uses. However, if you are looking for an equivalent of private IPv4 adresses you should take a look in RFC4193.

There you can see that fc00::/7 is the suitable subnet for testing purposes.

On the other hand you might have gotten an IPv6 prefix from your ISP which you could break down for your needs.

  • Yes, we also considered the fc00:/7 (or more precise fd:00/8) as option. However, this means that we would need to use 40 randomly created bits + subnet to get a /64 prefix (according to the RFC). Using e.g. 0000... as "random" number would violate the RFC, using real random numbers would render the IPv6 addresses probably very hard to remember for students which is also not desired. – sliebald Sep 13 '17 at 10:29
  • ULA is not equivalent to RFC 1918 Private addressing. With IPv4 private addressing, it is assumed that multiple networks will use the same addressing, but ULA is designed to try to prevent that so that the odds of two networks using the same addressing are very, very small. – Ron Maupin Sep 13 '17 at 13:16
  • I would also advise to use one or multiple prefixes as per RFC4193 from fd::/8. In real life, students will also not encounter /48s that will be easy to remember. – Marc 'netztier' Luethi Sep 13 '17 at 17:16
  • Thats true, however there are many fX.../64 prefixes that are special in IPv6, so these also don't really help with this problem. – sliebald Sep 14 '17 at 7:43
  • The odds of any particular pair of networks using the same addresses is very small but the odds of their being two networks in the world with the same ULA prefix are actually very high. – Peter Green Sep 15 '17 at 14:41
19

The IPv6 documentation prefix (2001:db8:::/32) must be used ONLY for documentation purposes. It means written examples, diagrams, PPT presentations, Textbook explanations, etc.

This range shouldn't be used in practical networks.

There is a "private IP range" of fc00::/7 which should be used for device testing, demos, courses, etc. as per RFC4193

Unicast adresses in this range must be used in local networks without access to public internet.

  • Regarding the fc00:/7, see my comment to Sascha R. We internally also had some people argue that the network we use can also be seen as documentation (or at least it is used as practical part of the documentation/explanation of IPv6 to students). Just from the RFC I can't see anything defining that documentation is strictly written text. Do you have any source for this interpretation? – sliebald Sep 13 '17 at 10:34
  • @StefanLiebald In APNIC they state that The documentation prefix is the IPv6 address range that has been set aside for this purpose. – jcbermu Sep 13 '17 at 10:48
  • @ jcbermu Hey, they also state 'In technical books, articles, and training material, it is often necessary to show examples of IP addresses in use.'. From my point of view "training material" would perfectly fit what we plan to do. Training for me includes practical exercises (or even could be seen synonymous). – sliebald Sep 13 '17 at 11:36
8

Just from the RFC I can't see anything defining that documentation is strictly written text. Do you have any source for this interpretation?

According to the IANA IPv6 Special-Purpose Address Registry, the 2001:db8::/32 Documentation address range cannot be used in source or destination addressing, is not forwardable (routable) nor globally reachable, and is not reserved by any protocol. That makes the address range unsuitable for use on an actual network.

How, for example, would you use any addresses in that range if they are not allowed to be used as source or destination addresses in IP packets?

In fact, you may run across some devices that simply refuse to use that address range, and that would be perfectly within standards. After all, IANA does own all the IPv4 and IPv6 addressing.

  • Nitpicking: afaik, IANA doesn't own the legacy IPv4 networks allocated by the Central Internet Registry priorIANA existence. – JFL Sep 13 '17 at 14:19
  • 4
    @JFL, IANA does, which is how we ended up with the 172.16.0.0/12 Private address range. IANA asked the law firm that "owned" the address range for some of it for Private addressing. The law firm refused, so IANA simply took it. The resulting lawsuit established that IANA actually owns all IP addressing – Ron Maupin Sep 13 '17 at 14:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.