|
It’s no secret that the supply of IPv4 addresses, on which the Internet has been based since the dawn of digital time, is rapidly running out. The official replacement is much larger IPv6 addresses, but I can report from experience that the task of switching is not trivial, and for a long time there will be a lot of the net that’s only on IPv4. So once the initial supply of IPv4 addresses run out, and the only way to get some is to buy them from someone else, what will the market be like?
The IP address market is sort of like banking on a planet where there is only a single gold mine containing 4 billion ounces of gold, run by the central bank, and the planet experiences severe deflation as the bank runs out of gold and can’t issue any more coins. I suppose one could try to analogize NAT (using private addresses on small networks like the one in your house) to fractional banking. We’d apply Gresham’s law to private RFC 1918 addresses (“bad addresses drive out good”), but that’d be a stretch.
What the bank analogy doesn’t capture is the routing costs. The reason that pure IP address markets won’t work is the routing externalities. Every time someone breaks a block of IP addresses into multiple pieces, that imposes costs on all the operators of the large routers that need to keep a route to every chunk of active IP addresses. But there is no workable way to impose per-route charges on end networks, nor to arrange to pay the money to the router operators to subsidize the costs that all the routes impose.
I suppose it might be possible for network peering arrangements to take into account the number of routes used by each peer as well as what they currently do, the amount of traffic in each direction. I’m not aware of anyone doing this, but I don’t claim to be totally au courant on peering negotiations.
Sponsored byDNIB.com
Sponsored byIPv4.Global
Sponsored byVerisign
Sponsored byWhoisXML API
Sponsored byCSC
Sponsored byVerisign
Sponsored byRadix
Thanks for helping to raise awareness about this very interesting point of symmetry John. For those who find this idea intriguing, I’ve added a couple of links to the original sources for these insights below.
To reduce that risk that anyone else will suffer a similar failure of imagination regarding the parallel monetary/banking phenomena corresponding to route de-aggregation, I’ll just offer a few basic observations here. The “routing externalities” that John refers to are a product of numerous factors, but by far the most important drivers in today’s Internet are (a) the absolute number of independent decision makers (a.k.a. “routing service providers”) that participate in the routing system, and (b), the varied, individually-defined, privately motivated strategies that those routing system participants adopt to advance their diverse local or private objectives.
It’s fair to say that any routing service provider that employs a routing strategy that involves thousands of separate independent variables (each producing an additional external routing table entry) is doing so because they believe that that strategy will more privately advantageous to them than adopting a simpler alternative strategy. Unfortunately, the shared routing system that ties the Internet together has a finite “carrying capacity” to accommodate such independent variables at any point in time. As a result, as additional routing service providers join the system, and incumbent service providers impose additional “discretionary” burdens on the system, the cost of participating in that system rises for all participants—in other words, routine system-driven “inflation” sets in. At the theoretical extreme, participation might not be possible at any price, e.g., if the aggregate route processing burden came to exceed the capabilities of even the most advanced and expensive routers of the day—and the Internet would be abandoned in favor of something better, or simply collapse.
So to recap:
More service providers > increased burden on the system’s finite carrying capacity.
Increasing use of “discretionary” strategies to maximize private gains > increased burden on the system’s finite carrying capacity.
As the aggregate burden approaches/exceeds the shared system’s finite carrying capacity > costs of participation rises for everyone > widespread system withdrawal in favor or something that is less expensive, or total system collapse
Now, for those who understand the role that banks and the act of “lending” play in the monetary system (and esp. for those who also understand the presumptive relationship between a bank’s “capital reserves” and the volume of it’s lending activities), this story should sound very, very familiar. The mechanisms outlined above are basically the same mechanisms that trigger monetary inflation—or in its most extreme manifestations, hyperinflation and monetary system abandonment of various forms (e.g., “dollarization,” reversion to barter, etc.). The absence in banking of anything that literally parallels “routing statements” doesn’t mean that one bank’s private, strategically motivated activities do not and cannot impose collateral costs on other financial service providers; assuming that it does is simply wrong.
Of course, those who believe that banks are “highly regulated” may claim that the root causes behind Internet routing system inflation and monetary system inflation could not be similar in any way. However, this is not really a counterargument because the parallels described above are completely independent of and unrelated to “root causes”—i.e., the resource flows, bilateral interactions, and mixed private incentives that connect banks to one another, and the identical linkages between routing service providers are still symmetrical regardless of whether and how one believes that those incentives have been shaped by external factors (e.g. deficiencies in TCP/IP protocols, failures of bank regulators, etc.).
Additional insights on these and other surprising points of symmetry between these two industrial sectors can be found online at:
http://labs.ripe.net/Members/tvest/industry-comparison-isp-and-financial-industry
and also at:
http://www.eyeconomics.com
TV