Log in

No account? Create an account

Previous Entry | Next Entry

IPv4 address space – what to do?

Another disturbing issue in the IPv6 migration is how IPv4 address space is currently managed. Some organizations don't manage it very well; they have many separate address blocks that are sparsely populated. Other address blocks haven't been allocated yet.

There are some address blocks that have been in existence and operational since the very early days of the Internet. The holders of these legacy address blocks are not all members of RIRs (Regional Internet Registries), which has caused a stir among the RIRs. Concern on the part of the RIRs is that they expend resources maintaining contact information, etc. for these networks. OTOH, the legacy block holders don't necessarily see eye-to-eye with the RIRs; they feel they don't benefit in particular from RIR services, or don't want to join an RIR and possibly risk losing their blocks. There actually was quite a debate on what to do about the legacy address space not too long ago; certainly, there will be more debate as the pool of available IPv4 address blocks shrinks.

One of the oldest legacy address blocks is held by MIT; it was operational even before the NCP to TCP/IP split. I remember when I started my UROP, 25 years ago, getting accounts on "detergent" machines in the "vax farm" (e.g. mit-borax, mit-mrclean). This was back in the days before domain names, so if they did not appear in the hosts.txt file, you had to refer to them by their IP addresses, which were of the form 18.x.y.z. Those were among the first "dotted quad" addresses I remember using.

As it turns out this block dates back even earlier – to a time when there were only 256 allocated networks. (This is because a previous implementation of IP was in use that did not divide the lower 24 bits of an IP address into either "classes", or the CIDR blocks of today.) I found the earliest reference to this block in IEN 91, under the name LCSNET. (LCS was the former name of the "computer science" part of CSAIL.)

So why is this network significant? Aside from the fact that it's one of the oldest operational IP networks, it's the only one of its kind that's solely allocated to a university. This is a matter of concern for organizations that haven't been able to get sufficient address space (if they've been able to get any). In the past, some other universities, such as Stanford, had legacy address space of this type (they were network number 36), but released it to the public pool. (It's currently reserved.) I'd heard at one point that MIT was thinking of switching to IPv6, but haven't heard anything lately.

Since IPv6 adoption is moving very slowly, opinions are mixed on what to do about the legacy address space. Some feel that IPv6 adoption will quickly progress to the point where the legacy space is irrelevant. Others are concerned that as IPv4 address blocks become scarce, a market will develop for them, and prices will rise. This could put pressure on MIT to release its block.