--- begin forwarded text X-Sender: rodney@module-one.tillerman.nu Date: Thu, 24 Sep 1998 07:32:20 -0400 To: dcsb@ai.mit.edu From: Rodney Thayer <rodney@tillerman.nu> Subject: hurricanes and digital commerce and financial cryptography Mime-Version: 1.0 Sender: bounce-dcsb@ai.mit.edu Precedence: bulk Reply-To: Rodney Thayer <rodney@tillerman.nu> Status: U Ok, Anguilla has had it's liter of rainwater worth of fame from this damn hurricane. Time for South Florida. So, ahem,... It's Thursday and there's a hurricane coming. There's plenty of beer in the starboard refrigerator and plenty of frozen White Castle hamburgers in the port refrigerator. The gas generator is all set to keep the server and the notebook PC battery chargers running. The local badly run telephone monopoly claims it can keep the telephone network up, we'll see if they are telling the truth when there's a foot of water on the road. In spit of the crypto-refugee claims, there's plenty of crypto work here, and this 'offshore' location is only about 100 feet from the mainland so it's not actually exporting code when I drive back to the office from the local computer store with the PC in my car. Here's my commerce question. How often do people expect to cycle private keys from PKI systems? I mean, suppose you have a web server, or an IPSec router, or some related device. I would assume you'd actually generate a new key pair every time you get a new certificate. So, for example, if you get a new certificate every six months for a web server, you would use a new private key. It turns out there is some debate over this. Some people assume the private key lives essentially forever, and that you'd simply get a new certificate against the same old key pair. I myself think this is rock stupid because one of the threat models is that the private key gets stolen, and the longer it's used the longer it's stealable. This relates to hurricanes. If you know any IPSec-aware routers, or any secure web servers, or any other PKI-based devices, and they are in the Florida Keys, you had better assume they'll need new key pairs as of some time Sunday. You see, unless I've missed the "waterproofing" requirements in FIPS 140 (the government security hardware standard), then those boxes will all be under water and will require replacement. So any of these PKI Service Provider models that assume the private key is immortal are about to be proven wrong. And there better be something in your revocation procedure for "the private key was destroyed by an act of nature". Look that up in your CPS and your Digital Signature legislation. But I need to go prepare, it's supposed to start getting exciting here in Clearwater, some time today. [I pass the torch to some DCSB subscriber in the New Orleans area...] -------- Rodney Thayer <rodney@tillerman.nu> "layer of sand separate transmitted and received packets". (with apologies to Keith Law son.) For help on using this list (especially unsubscribing), send a message to "dcsb-request@ai.mit.edu" with one line of text: "help". --- end forwarded text ----------------- Robert A. Hettinga <mailto: rah@philodox.com> Philodox Financial Technology Evangelism <http://www.philodox.com/> 44 Farquhar Street, Boston, MA 02131 USA "... however it may deserve respect for its usefulness and antiquity, [predicting the end of the world] has not been found agreeable to experience." -- Edward Gibbon, 'Decline and Fall of the Roman Empire'