North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

History of 24/8 (was RE: for folks tracking DDOS sources or reading the GRC attack log)

  • From: David R. Conrad
  • Date: Tue Jun 26 01:05:28 2001


[Long, boring, and not much operational content. You are warned]

A bit of biased history:

Long ago, in an Internet very different than today's, Jon Postel was convinced by people with more than a little knowledge about such things that this "IP over Cable TV infrastructure" idea might actually take off. In fact, he was convinced that it was going to work so well that all the Cable TV providers all over the world were going to besiege the IANA like a biblical swarm of locusts and Jon would spend all his time fending off requests for huge blocks of addresses.

My understanding of why things went to the IANA was that @Home was rejected for a large block of address space and they appealed to the IANA (as was the policy in those days). After much discussion, multiple requests for more information, proof, first-borns, etc., Jon agreed that the IP over CATV infrastructure model of operation was sufficiently different that it justified special treatment (the argument being, to my understanding, that IP over CATV provided something new at the time: always on end user service).

The convincer (I'm told) suggested that the Cable TV industry was going to form a trade association that would handle industry coordination requirements like, oh say, allocation of IP addresses. Jon decided to allocate 24/8 to this (at that time not-yet-existant organization) with the understanding that all IP over CATV infrastructure address requirements were to be managed by this new trade association once it was formed. Until it was formed, the IANA would authorize ARIN (InterNIC at the time, I believe) to allocate address space out of the 24/8 block. Of course, history has shown that no such organization was formed (if you assume AT&T is not that organization).

Anyhow, when APNIC, InterNIC, and RIPE-NCC found out about all this, we all expressed significant unhappiness as we didn't think there was any significant difference between a IP over CATV provider and any other ISP. However, what was done was done and it would've been a bit difficult to get @Home to return the block since they'd already allocated a bunch of customers to it. APNIC gave IP over CATV ISPs the option of allocating out of normal APNIC space or forwarding their request to the IANA. I think one request was forwarded to the IANA for an IP over CATV provider (but my memory is hazy). Not sure if RIPE-NCC forwarded any requests. After a while, I saw 24/8 as just another InterNIC/ARIN block, albeit one with a weird genealogy.

At the time, 24/8 was pretty controversial. Many of the loons whinged and moaned about the unfairness of it all, making all sorts of bogus claims like @Home got the entirety of 24/8, that Jon was biased about the allocation since he worked with many of the @Home principals, etc. But then again, the loons whinge and moan about pretty much anything. Controversy flared up over this more recently when ICANN used 24/8 as an example of RIR-independent address space that would justify ICANN doing allocations of address space themselves. Suffice it to say that this was a poor example.

Now, isn't that more than you ever wanted to know?

Rgds,
-drc
(who played a small role at APNIC in a previous life)


At 02:24 PM 6/26/2001 +1000, Bruce Campbell wrote:
On Mon, 25 Jun 2001, Richard A. Steenbergen wrote:

> On Mon, Jun 25, 2001 at 04:54:12PM -0700, Mike Batchelor wrote:
> >
> > >     24.0/8 is the "cable block".
> >
> > No it's not.  Check out 24.132/14 for instance.

and 24.192/14

> IANA and ARIN seem to think it is... Check out:
> http://www.iana.org/assignments/network-24

24/8 has been used by all 3 RIRs to make assignments to cable modems in
times past.

Recent events have lead to the non-ARIN parts of 24/8 being taken up by
ARIN, in addition to the non-allocated-to-a-RIR parts of 24/8.  Updates to
various databases will simply take a bit of time, thats all.

--==--
Bruce.