3

can someone explain to me why we cannot use the class E IP space? I know it's reserved but this is a huge chunk of IP addresses they could make available. Would this bring such huge issues that it cannot be done? Are these addresses hard coded into system stacks as non-routable ? I know there were some projects to get it done but none of them was successful so I just want to know, are the issues of getting it done so much bigger than the gain of having them free to use? And what are the issues exactly? Is it because it's not worth it as we should migrate to v6 completely?

5 Answers 5

7

Are these addresses hard coded into system stacks as non-routable

Not so much non-routable, but "invalid". There are a great many systems across the internet with that logic builtin. (and some that mistakenly treat 240/4 as multicast!) It was decided long ago to not bother with promoting "Class E" to usable address space -- the effort would be almost impossible, and a /4 would only prolong the inevitable by months. In other words, GET ON WITH IPv6 ALREADY.

(Just look around your own networks. How many legacy, unsupported devices are still there? Print servers, door bells, industrial controls (HVAC, elevators, etc.), phones, consoles, etc, etc, etc. Not to mention the old but still 100% functional routers and switches - who definitely must support this.)

4
  • +1 for "and a /4 would only prolong the inevitable by months. In other words, GET ON WITH IPv6 ALREADY."
    – tjd
    Commented Oct 5, 2023 at 17:20
  • It's kindof insane that something so futuristic as air conditioning ON THE INTERNET!!! was built with pretty-much obsolete Internet when they installed it. It was normal to use v4 a couple of decades ago but even then people knew it wouldn't last out long enough for the expansion that was inevitable. So instead there's ridiculously complex routing and virtual addresses, with NAT semi-smartly using socket numbers as a pseudo-IP. All that money and effort, gone wrong for the lack of 10% extra effort, instead we have to make 50% more just to kludge it.
    – Greenaum
    Commented Oct 5, 2023 at 19:44
  • 1
    Remember your history... IP (v4) was an experiment. It was never envisioned to be a global network protocol. No one at the time had any idea of what eventually became the internet, or that it would be such a utility.
    – Ricky
    Commented Oct 5, 2023 at 19:47
  • @tjd And today, it wouldn't last a month.
    – Ricky
    Commented Oct 5, 2023 at 19:49
5

The address space is likely hard-coded in various systems as non-usable, yes.

Having it usable globally would would in practice require getting all operating systems, routers and firewalls accept them, and in addition would likely require many individual network and system administrators to revisit their manually-set filters dropping that and other such reserved blocks. There would be hold-outs, there would be issues with specialist and embedded equipment, and in the meanwhile, services trying to run on addresses from that address space would be mysteriously unusable by some end-users. Not good if one happens to be trying to run a business.

It's not like all systems prevent the use of those addresses, however. To pick anecdotal data from what I have at hand, a recent Linux allows assigning an address from 240/4 just fine, and on Junos the addresses can be used if you explicitly tell the system to accept them with special configuration. Then,

testlab1# run ping 245.6.7.2
PING 245.6.7.2 (245.6.7.2): 56 data bytes
64 bytes from 245.6.7.2: icmp_seq=0 ttl=64 time=3.002 ms
64 bytes from 245.6.7.2: icmp_seq=1 ttl=64 time=3.389 ms

There appear to be indications that some rather big players have made the space usable within their networks. The RIPE Labs article "240/4 As Seen by RIPE Atlas" mentions their network probes seeing addresses from 240/4 in traceroutes and there's at least one earlier rumor on that, too.

Incidentally, that email thread mentions that it took ARIN about 3-4 months to distribute a /8 to its members in 2011. At that speed, a /4 would last globally for something like a year, so in addition to the likely significant trouble, there would also be relatively little benefit from making the address space usable.

2
  • 1
    Not "likely"... "absolutely is". No Cisco device ever made will allow 240/4. They won't even route it. The only way to make them globally usable is by inventing a time machine, going back about 40yrs, and never marking them as unusable. (ARIN was handing out more than a /8 per month.) Yes, there are those that have plumbed 240/4 into their own internal networks, but they have to be very careful because you never know what things will accept it, and what will mistake it for multicast.
    – Ricky
    Commented Oct 4, 2023 at 20:13
  • 1
    Not just Cisco; Windows also refuses to talk to 240/4 entirely. (Though, updates could be made available to both Windows and IOS, I doubt it's hardwired even in routers, but that's highly unlikely at this point.) Commented Oct 5, 2023 at 6:14
3

First, network address classes are dead (please let them rest in peace), killed in 1993 (two years before the commercial Internet in 1995) by RFCs 1517, 1518, and 1519, which defined CIDR (Classless Inter-Domain Routing). We have not had network address classes in this century.

IPv4 addresses with the first four bits set to 1111 are Reserved, and that reservation is built into IPv4 itself, as explained by the IANA IPv4 Special-Purpose Address Registry.

It would be prohibitively expensive to retrofit every IPv4 device in the world to recognize such addressing, and it really does not gain all that much (we would still not be able to give but a fraction of the extant IP devices a unique address). IPv6 has solved the IP address shortage, and it has restored the IP end-to-end paradigm and improved IP over IPv4.

2
  1. It has become clear that people will not transition away from IPv4 until/unless continuing to use IPv4 becomes too expensive. People have been toying with IPv6 for years but few operators got serious about it until after the RIRs announced exhaustion. Even a decade after said exhaustion less than half of clients can access IPv6 servers.
  2. Many systems outright refuse to use class E addresses. Even if a standard was published saying such addresses should be treated as regular unicast, it would likely take many years to phase out the old systems.
  3. The allocation rate in the years leading up to exhaustion was pretty high and was accelerating. The "Class E" space represents 16 /8s. Even if it had been used as public space it would have only delayed exhaustion by about a year if that.

Put these things together and allocating the space for public use was a non-starter. Using them for private use was more plausible, but getting to the point where such addresses could reasonably be used on general purpose networks would still be an uphill battle that might distract people from transitioning to IPv6.

2
  • 1
    Not "years", but decades. (like 3 decades) And there are still plenty ISP that still have zero IPv6 anywhere. None of the enterprise connections I've had in the last 20 years supported v6. (In AT&T's words, we weren't a big enough operation to be connected to that router.) Earthlink still doesn't - despite dozens of v6 blocks. [the only v6 they ever announced was part of OCCAID - an experiment from a single employee.]
    – Ricky
    Commented Oct 4, 2023 at 20:03
  • Addendum to point 1 - or that the RIRs disallow any kind of transfer of addresses & gradually reclaim those that are currently issued in reverse order in which they were issued.
    – Bib
    Commented Oct 4, 2023 at 20:39
0

why we cannot use the class E IP space?

In the local network, you could do it. And I've heared that there are hosters who already do that!

And in the Internet?

If it is true that hosters or ISPs use "class E" internally, the corresponding companies will refuse to support "class E" addresses in the Internet.

It will neither be possible to access web sites hosted there from a computer that has a "class E" address, nor it will be possible to access a "class E" web site from a computer connected to such an ISP.

Router operators who have replaced their IPv4-only routers by IPv4+v6 routers will probably also refuse to replace their routers again if the new ones don't support "class E".

And router operators who refused introduction of IPv6 up to now will also not see any need of introducing "class E".

So the introduction of "class E" in the Internet would not be faster than the introduction of IPv6.

And using "class E" in an Internet where many routers don't support "class E" would cause many problems that IPv6 did not have in its first days:

Just think what happens if a "class E" computer tries to connect to some web site and the routers in between don't support "class E"...

It would take decades until "class E" could be used without problems in the Internet ... if the router operators will still support IPv4 then!

... this is a huge chunk of IP addresses ...

If the demand in Asia has not changed in the last years, companies there are waiting for 920 million IPv4 addresses.

"Class E" is only 268 million addresses. The "class E" addresses would not even be enough to supply all the companies in Asia waiting for addresses just now!

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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