Everyone knows that "it's a small world, after all."
The only problem with that Disney-esque bit of common knowledge is that "everyone" happens to be dead wrong.
As G. Harry Stine pointed out in a 20-year-old Analog editorial, the small world shibboleth is most likely a product of the tunnel vision that afflicts urbanites the world over. For city-dwellers, everything is just a few miles away. Jet aircraft that fly from coast to coast or continent to continent at 35,000 feet -- well above the cloud deck -- add to the illusion.
You drive to the airport, get on a plane, spend a few hours flying above a uniform mass of fluffy, white clouds, get off the plane and, presto, you're at another airport, just a few minutes' drive from a city much like the one from which you departed. You experience none of the landscape between the two aerodromes, have no way to appreciate the often-vast spaces between them and wind up with the unavoidable impression that the distance between one place and the other must therefor be fairly trivial.
I'm here to tell you that it just ain't so. In case you missed the memo, the globe isn't all that darned small. In fact, it's actually a great, big, complicated world -- and it's getting bigger and more complex all the time.
Granted, the amount of physical land area isn't growing -- in fact, thanks to the now-undisputed effects of global warming, it's actually shrinking. However, the population of the planet is inexorably increasing -- and the labyrinthine set of concomitant problems that accompanies that expansion is burgeoning right along with it.
Take what's already happened to the Internet as an example. In the last seven years, the total number of connected nodes has exploded from less than a million in toto to well over 100 million. Sometime in the next five years, it will pass the billion mark -- and keep right on going.
Something needs to be done -- and soon -- to cope with the tidal wave of technical problems the flood of new nodes has already created and will continue to to spawn until the entire planet is connected. I'm not just talking about the problem of address space depletion -- although nanotechnology will add to that headache before too long. It's the cancerous hypertrophy of our router tables, the fundamental insecurity of our packets, the lack of easy extensibility of our core networking protocol and the continuous proliferation of new requirements for it to fulfill.
In short, the fly in our collective ointment is the increasingly-evident creakiness of IPv4 -- a technology now nearly old enough to run for the House of Representatives.
I Want a New Drug
Think for a moment about the shape of the world that spawned the old standby. Back when TCP/IP was first developed, the dominant computing paradigm -- at least in the academic universe in which the ARPAnet incubated -- was host-centric, based on dumb terminal access to minicomputers. It was the golden age of DEC, when VAXen stalked the Earth and there were no computer "users" -- there were only the programming elect and the data-entry hewers of wood and drawers of water.
That there might, eventually, be as many as tens of thousands of nodes on the network meant that the classful IP addressing scheme was ridiculously overdesigned. Why, it actually made possible literally millions of networks! Sixteen or so, in fact.
Then came the 1990's and the Internet population explosion -- not to mention the dawning of the Age of Users, who neither toiled at data entry nor administered, nor yet hacked code, but who merely surfed the digital seas.
And still, old faithful IPv4 diligently connected us all.
There was the nasty problem of address depletion, of course, but the development of Classless InterDomain Routing, (CIDR), Network Address Translation, (NAT) and Border Gateway Protocol version 4 (BGP-4) took care of that -- or at least put it off for a good, long time to come.
Well, those Band-Aids were first applied six years ago and they are now all scuffed and dirty -- and soon they will fall right off the unscabbed wound. In the meantime, a whole new host of issues has arisen: security issues, transparent addressing issues, route flapping, stepped quality of service requirements and the rapid convergence of voice, packet data networking and wireless and other appliance connectivity.
IPv4 addresses none of these well -- and the address depletion problem will grow geometrically as new nodes flood onto the network over the next few years. NAT cannot meet the challenge -- expecially if Peer-to-Peer becomes as central a technology as the computer trade press would like us all to believe will happen -- because the sheer demand for publically-visible adresses will eventually exceed the maximum theoretical supply.
The solution to a lot of these problems lies in weaning ourselves off the old, familiar IPv4 and migrating the entire Internet -- and all the networks that connect to it via TCP/IP -- to IPv6, the anointed successor protocol. That's a goal that's been just around the corner for a long time, now.
And, unfortunately, it looks very much as if it will continue to be just around the corner for a good, long time to come.
That's too bad, because IPv6 offers a lot of networking goodies: an unimaginably vast address space, for one thing, (there are exactly 340,282,366,920,938,463,463,374,607,431,768,211,456 available node numbers in IPv6's 128-bit address space -- enough to provide a separate address for every water molecule in every ocean on the planet, with plenty left over to address every molecule in every human, cat, dog, goldfish and guinea pig now alive.) Mandatory encryption of packet contents, courtesy of IPsec, for another. Auto-magic configuration -- and equally invisible auto-renumbering -- for a third.
And then there's that nifty extensible header feature that should allow IPv6 to gain nearly any conceivable functionality future needs may demand without requiring any changes whatsoever to the core protocol.
The heart of the migration problem lies at the top three layers of the protocol stack, of course.
The three layers to which I refer are the ones Carl Malamud defined as extensions to the ISO 7-layer stack model: religion, politics and money. And all three have played a key part in the agonizingly-slow adoption of the once-and-future IPng.
Afflicting the Comfortable
Cisco and Microsoft have been the key foot-draggers on the technology side. Both malefactors have had ample motive for malingering on monetary grounds alone -- Cisco is, for all practical purposes, a monopoly player in the Internet core router game. As its non-stop acquisistion binge of the past several years demonstrates, there's a lot of money at stake.
And a changeover to IPv6 might very well bring that happy circumstance to an end.
Paradigm shifts are like that. They open previously-closed doors to the entry of new players and/or the accession to dominance of formerly-minor ones. And, when you are the house dealer for the only game in town, they are always a threat.
Microsoft's reasons are equally easy to fathom. Their intial foray into IPv4 occurred only hesitantly and with multiple missteps along the way and their current implementation is still painfully bug-ridden. Moving to IPv6 will garner them yet another set of lumps and scratches -- and entail development costs that they'd rather not have to expend.
For the Redmond behemoth, there's no immediate profit in IPv6, you see -- and, just as importantly, no competitor whose ace requires trumping. So it can wait.
Now it's true that both of our foot-dragging friends have seen the light and formally sworn allegiance to the IPv6 flag -- at least for publicity purposes. But Cisco's stated implementation timetable has already slipped -- as of this writing, only their beta IOS software appears to support Phase I functionality. And the entire tone of their public statements on the subject is, at the very best, reluctant.
Their committment appears to have been made not so much "kicking and screaming" as "sulking and whimpering."
You can load an IPv6 stack and, by taking the performance hit that encapsulating its packets in old-school IPv4 wrappers entails, actually access the so-called 6bone. It'll give you a taste of what may yet come to be and help you familiarize yourself with what wants desperately to become the Next Big Thing in Internet technology.
Hey, it's a big, 'ol world out there. And you -- and your users -- are going to have to live in it as long as you stick around.
And, if you doubt that it's all that large a place, next time you have to get across the continent, try walking to and from the airport. Better yet, get a horse and try riding coast-to-coast.
You might be surprised at how big a world it is -- after all.
(Copyright© 2001 by Thom Stark--all rights reserved)