Slashdot Mirror
RSA-640 Factored
gslin writes to tell us MathWorld News is reporting that RSA-640 has been factored. F. Bahr, M. Boehm, J. Franke, and T. Kleinjung, memebers of the German Federal Agency for Information Technology Security (BSI) announced they had cracked the 193-digit number last Friday using the General Number Field Sieve. The team purportedly used 80 opteron CPUs and 5 months to achieve victory.
-Not always true. Say I can come up with a 2048 bit encryption, that is just increase the key size from 256 to 2048, I can to that in a second. It is going to take _a lot more time_ for the computing power to overcome that increase.
If quantum computing will come around, I'll just switch to quantum encryption. Then you'll have to break the laws of QM to "break" the scheme. There are already rudimentary quantum encryption devices but there are no quantum computers that can take on even a 64 bit key space.
The best bet instead of brute force is to do "human engineering" and look for other ways to obtain the information. The inherent math of the algorithm is rarely the weakest link, it is the people and then the particular implementations of the algorithms that are exploited the most.
However what if that team had decrypted a banks RSA key. Sure it may have taken 5 months and the bank may well have changed keys since. But what if they captured all the packets sent/recieved using the cracked key. I guarentee there would be a bit of useful information in them. I know I don't change account numbers / credit card numbers every 5 months.
From a security point of view, "less than five seconds on modern hardware" is absolutely unnecessary. After a transmission has been intercepted, and we work under the assumption that essentially all of them are these days, that transmission has to remain secure potentially for a very, very long time (sure, an encrypted radio communication might have no operational significance in two hours -- an encrypted dossier of a field agent had better stay secure until he's dead, and an encrypted report on security breaches in US nuclear command&control protocols better stay unbroken for the better part of the next century). And what takes a beowulf cluster of supercomputers 5 months to do today will be possible to crack on a botneck in less than a week by 2010, I guarantee you (I write distributed applications for a living and worked with a three-letter agency which was rather miffed that a competing TLA had the existence of their own internal distributed cracker leaked to the press back in 2002 or so). A couple years after that it will be a few minutes on dedicated hardware and a few years after that a desktop machine will smash it as a matter of course. Events like this are warning flares to people with serious security needs that you need to start transitioning to harder codes or longer key lengths.
Help poke pirates in the eyepatch, arr.