Home / Blogs

Map of the Internet: The IPv4 Space of 2006

An intersecting representation of the IP Address space on a plane using a fractal mapping that preserves grouping… Section of the map also shows the blocks sold directly to corporations and governments in the 1990’s before the RIRs took over allocation.

Created by Randall Munroe of xkcd.com (Source Permalink)

Editor’s Note (Dec 28, 2006): Please note that the IP Address illustration shown above is comical in nature and should not be mistaken for a complete or authoritative representation of the IPv4 Address space.

By CircleID Reporter

CircleID’s internal staff reporting on news tips and developing stories. Do you have information the professional Internet community should be aware of? Contact us.

Visit Page

Filed Under


Matthew Elvey  –  Dec 26, 2006 9:55 AM was “sold” to GE?  Domains were once free; IPs were too, or is my memory fuzzy?

Gregory Wexler  –  Dec 27, 2006 6:12 PM

I wonder why “216” has text in it that says “something aweful” ?

Tom Vest  –  Dec 28, 2006 3:31 AM

There is already too much misinformation and confusion about IPv4 addressing; please do not compound the problem by redistributing this terrific, albeit somewhat inacurate and misleading *comic*, as if it were an authoritative source of information on this subject. Note the source, “xkcd.com: A webcomic of romance, sarcasm, math, and language”. xkcd is indeed a consistently entertaining (and often very insightful) content source on the enumerated subjects. However, the artist’s description (repeated above) of early classfull allocations as “blocks sold directly to corporations and governments” clearly marks this particular subject as outside the authors’ sphere of expertise.

Nothing wrong with having fun with obscure and/or serious subjects… but retelling a joke as if it were a true story is neither illuminating nor funny…

The Famous Brett Watson  –  Dec 28, 2006 3:15 PM

I don’t care if it’s produced by a webcomic author: the fractal layout is a cool idea, and the only remaining question of any relevance is that of accuracy. (I’ll find an answer to that question if I need to rely on this data. As it happens, I don’t.)

Tom Vest  –  Dec 28, 2006 4:24 PM

I agree that the fractal layout is a cool idea, but it seems to me that the way it was initially re-presented without comment here, in what is generally regarded as a “serious” Internet policy forum, warranted a clarification. Your tendency to find confirming references outside of CircleID is not universally shared, as I have seen CircleID postings cited in academic papers more than once, and used out of context (e.g., to imply the opposite of the cited author/article) more than once. Better to provide full attribution and context up front, than to have to address cascading misunderstandings down the line.

Graphics and other non-visual rhetorical tropes are esp. powerful, and therefore important, in obscure subject areas like this.  I would think that a good benchmark for any visualization scheme is whether or not it “illuminates” or “clarifies” or at least “faithfully represents” the subject matter. Cool is also a virtue, but shouldn’t be elevated above the others, IMO—otherwise we risk getting stuck with “a series of tubes” or something equally unhelpful…

How about this for an alternate visualization? The entire 4.2b IPv4 address span arrayed as a flat circle—or rather a pie, composed of 256 equal slices. Each of these slices represents one of the /8s depicted in the original xkcd map. Look a little closer and you see the pie is actually a bullseye, composed of three nested circles. The center point represents—the default route (not usefully represented in the xkcd version). Each complete slice extending from the center to the outermost perimeter emcompasses a full /8 (16.8m IP addresses), as in the xkcd graphic. Look closer still and you find that the middle segment of each slice is composed of 256 equally sized radial segments, each representing one of the /16 prefixes (65,536 IP addresses) encompassed within or “covered” by the associated /8. Look even closer and you note that each of those (256*256 = 65,536) intermediate radial segments corresponds to yet another 256 equal segments in the outermost ring of the bullseye. Each of those outer segments touches the ring’s outer perimeter, and encompasses 256 individual public IPv4 addresses. Along the surface of the IPv4 prefix-pie, you find each of the 4.2b public IPv4 addresses—the proverbial ends of the e2e Internet—but with all of their routing and/or allocation-relevant associations intact.

If I understand it correctly, the suggested IPv4 prefix bullseye-pie “preserves grouping” ala the xkcd graphic, but does so even more thoroughly, and in a way that could be labeled in more acurate and informative ways than the original. Such a visualization would be equally useful to understanding current/evolving Internet production (i.e., routing—with segments labeled by originating institution or country rather than region of allocation) as to cumulative IPv4 address allocation history (the implicit subject of the xkcd version).

Admin  –  Dec 28, 2006 6:59 PM

An editor’s note has been added to the post to emphasize the non-serious nature of the illustration. Your feedback is much appreciated.

The Famous Brett Watson  –  Dec 29, 2006 3:37 AM

Tom, it seems to me that the most important thing about this representation is the allocated vs unallocated difference—that is the areas which are green relative to those which are not. The top right-hand corner of the map represents IP-space which is “reserved” (and informally known as “Martian addresses”) if I recall correctly, so although it is unallocated, it doesn’t really represent a portion of IP space that could easily be allocated. Even so, there is a lot more green area on this map than I would have supposed, given the usual dire warnings about IPv4 address scarcity.

Is anyone willing to step up to the plate and vouch for the extent to which this data is accurate? I assume the primary source is WHOIS—I know of no other public sources.

As to the matter of your alternate representation, it’s not the representation of the data that’s the key point here, but rather the accuracy of the data. Even so, I note a problem with your pie-chart approach: it’s difficult to maintain a constant area per address when mapping onto a circle.

Also, while I’m posing, here’s an answer to Gregory Wexler’s question about “something awful”. The map pinpoints the IP address location of a number of infamous websites including Google, Digg, Flickr, Slashdot, eBay, and CraigsList (all situated near the middle of the left edge). “Something Awful” is just another of these infamous sites—and you can go find it at your own risk.

Tom Vest  –  Dec 29, 2006 5:38 AM

You ask: Is anyone willing to step up to the plate and vouch for the extent to which this data is accurate? I assume the primary source is WHOIS—I know of no other public sources.

Actually, whois has nothing (or at least nothing consistent) to do with the xkcd map. It was almost certainly taken from this high-level view of bulk address space delegations from IANA to Regional Internet Registries (RIRs), or in a few (mostly historical) cases, to what would now be called Local Internet Registries (LIRs). The authoritative source for the finer grained, cumulative allocation view of IPv4 addresses that I proposed would be the “delegated-

” files, which are published online in a standardized format every day by each of the Regional Internet Registries. They are easy to find online. The “delegated” files provide the chronology and national provenance of the individual prefix level transactions by which exclusive but conditional/temporary beneficial control of IP number resources passes from RIRs to LIRs (enterprise-supporting ISPs) and individual networked institutions. Overall they encompass some 50k or so entries, as opposed to the 256 entries in the IANA file and the xkcd map.

Then you stated:

As to the matter of your alternate representation, it’s not the representation of the data that’s the key point here, but rather the accuracy of the data.

I have a hard time imagining anyone making such a statement about a subject that they actually care about. I suspect from your other postings here that you’re very interested in DNS. Would you agree that the statement “domain names possess some value-relevant features, and are subject to some rules regarding their beneficial control” is accurate? Does that accuracy confer any utility or other kind of merit upon the assertion? Does it warrant any/every conceivable conclusion that might follow, provided only that the “accuracy” is preserved? Your own writings suggest that you are quite willing to distinguish between good and less-good, or useful and less useful   analogies, etc. I am just doing the same thing here for a subject that I care about.

Then you commented:

Even so, I note a problem with your pie-chart approach: it’s difficult to maintain a constant area per address when mapping onto a circle.

I’m no mathematician (!), but in the alternate mapping that I proposed, I believe the exact/constant area per IPv4 address would be (pi/2^31) radians—i.e., the length along the perimeter of a circle formed by 4,294,967,296 equidistant points, one for each theoretically possible IPv4 address. Each of the 256 /8 prefixes (or number segments) would cover (pi/2^23) radians, while /16 and /24 prefixes would cover (pi/2^11) and (pi/2^7) radians respectively. CIDR allocations are left as an exercise for the reader. The largest segments could be labeled just as in the xkcd map (less the spurious “sales” assertions). However this approach would also makes it possible to fill in those abiguous continental or “various registrars” regions with more detailed and useful data.

I’ll welcome math corrections, but otherwise feel like this is a good place (for me) to stop…

The Famous Brett Watson  –  Dec 29, 2006 5:08 PM

Tom, thanks for the pointer to the IANA high-level delegation file. A brief non-exhaustive comparison of this data with the diagram uncovers no obvious discrepancies. The latter does contain more detail than the former in some cases, such as the delegation of 126/8 to Japan, so it’s clearly not the only source, and a full verification isn’t possible without further investigation. My only nitpick with the map so far is that 10/8 is green, and I think it shouldn’t be, but it’s a special case.

Having said that, I’m a little confused as to the nature of your objections. Your initial complaint (comment #3) described it as “somewhat inaccurate and misleading”, followed by specific reference to the phrase “blocks sold directly to corporations and governments”. Granted, that sentence is somewhat inaccurate and misleading, but is that all? Your tone suggested that the problem was a little more pervasive than that.

Aside. To set the record straight on the allocation (not sale!) of Class A networks, note the following from RFC 1917, section 2.1. “Of the lower half of [Class A] address space, 22 [Class A networks] were assigned pre-1982, 6 were assigned between 1982 and 1987, 26 were assigned between 1988 and 1992, and 2 were assigned between 1993 and 1995.  In May of 1995 four Class A networks previously assigned have been returned to the IANA.”

In comment #5 you describe an alternate method for presenting the data, and your proposal seems to have as one of its goals an ability to represent all the data, rather than just the top level (or so) of delegation. This is reinforced by the first part of comment #8 where you speak of “the finer grained, cumulative allocation view of IPv4 addresses that I proposed”. Is the incompleteness of this map another of its “inaccurate and misleading” features? I think not: a map of the world is not expected to include detail down to the street level (and would be unreadable if it did). This map of IPv4 space is like a map of the world, showing the equivalent of country boundaries and major cities.

A fully zoomable map of IPv4 space and its delegations would be a very cool thing. This isn’t it. This is just a representation of the whole IPv4 space at a very high level, just like a map of the world. It even has crinkly edges to make the regions look a bit like land-masses. If you’re going to criticise it, criticise it for what it is and any inaccuracies it may hold at that level.

Alessandro Vesely  –  Dec 31, 2006 9:41 AM

That map is


Satirical is a better adjective than comical for describing how it points out there’s still grass in the IPv4 space.

By it own nature, that ”.com/c” also remarks that there is no such a thing as a Public Internet Map. Some sell ip-to-country obfuscated conversion tools. Postmasters would probably appreciate also country-to-ip-regions maps: when provided in a compact CIDR notation, they could be useful for composing country oriented SPF policies.

Comment Title:

  Notify me of follow-up comments

We encourage you to post comments and engage in discussions that advance this post through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can report it using the link at the end of each comment. Views expressed in the comments do not represent those of CircleID. For more information on our comment policy, see Codes of Conduct.

CircleID Newsletter The Weekly Wrap

More and more professionals are choosing to publish critical posts on CircleID from all corners of the Internet industry. If you find it hard to keep up daily, consider subscribing to our weekly digest. We will provide you a convenient summary report once a week sent directly to your inbox. It's a quick and easy read.

I make a point of reading CircleID. There is no getting around the utility of knowing what thoughtful people are thinking and saying about our industry.

Co-designer of the TCP/IP Protocols & the Architecture of the Internet



New TLDs

Sponsored byRadix


Sponsored byVerisign

Threat Intelligence

Sponsored byWhoisXML API

Domain Names

Sponsored byVerisign

IPv4 Markets

Sponsored byIPv4.Global

Brand Protection

Sponsored byCSC