The following text is copyright 2003 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

 

Miscounting and misunderstanding addresses

 

By Scott Bradner

 

The BBC reports that the end of the Internet world is near.  They are not the first to do so, nor will they be the last, but their story caught my eye the other day. (http://news.bbc.co.uk/2/hi/technology/3211035.stm)  This particular report, like far too many in the last few years, is about the Internet running out of IP address space real soon now.  They are wrong.  Not only are they wrong in the basic story, they are also wrong in almost all of the details they present.

 

IP addresses have a dual function in the Internet.  First they are used as location information to say where in the network infrastructure a particular computer happens to be.  Second they are used to identify the particular computer (actually a particular interface on a particular computer).  IP addresses can be public or private.  Private IP addresses are used within a network and must be translated to public addresses if the user needs to reach another site on the Internet.  Public IP addresses are assigned (not sold) by regional IP address registries.  There are currently four of these, each with its own geographic scope, ARIN (www.arin.net) which covers North America, parts of the Caribbean and Africa south of the equator, RIPE NCC which covers Europe and Africa north of the equator, APNIC which covers the Asia-Pacific region and LATNIC (www.lacnic.net), the newest registry, which covers Latin America.  A new registry is being formed that will be responsible for address assignments in Africa.

 

The BBC report did get it right that IPv4 has an address space of about 4 billion addresses but that is about where they stopped being right.  They say that the IPv4 address space will run out sometime in 2005, but the reality is that current projections do not show that it will run out, assuming the current rate of assignments, for almost two decades. (see http://www.potaroo.net/ispcolumn/2003-07-v4-address-lifetime/ale.html)  The BBC reported that IPv6 will provide 64 billion extra addresses where it's actually 340,282,366,920,938,463,463,374,607,427 billion extra.

 

The BBC also reports that: "the global distribution of available IP addresses is extremely unbalanced.  Most of the numbers remain in the USA, where the technology was originally invented."  This is misleading at best.  Since addresses are assigned where the need is it is far from surprising that a lot of addresses were assigned where the technology was actually used.  But there is no global preassignment of addresses, the registries get whatever addresses they need when they need them and APNIC assigned more addresses than any other registry in the last two years. (See http://www.arin.net/statistics/jointstatsoct03.pdf.)

 

As one of the managers, along with Allsion Mankin, of the IETF project that resulted in IPv6 I do think that something like IPv6, and I expect it will be IPv6 itself, will be needed over the next decade as the Internet expands to cover many more applications such as IP-based cell phones.  But there is no reason to panic.  IPv6 is well along in deployment and will be there when we need it.

 

There may be reason to worry that the poor coverage of this type of issue by the popular press causing panic among their readers.  But clueless reporting is hard to prevent, even when the facts are readily available to anyone who asks before assuming they know it all.

 

disclaimer: As a member of the ARIN board of trustees I'm biased towards truth about IP address assignments policies but I did not ask anyone there or at Harvard, where, of course, cluelessness is never to be found, about this column.