Cracking Passwords in the Cloud: Breaking PGP on EC2 with EDPR(news.electricalchemy.net)
news.electricalchemy.net
Cracking Passwords in the Cloud: Breaking PGP on EC2 with EDPR
http://news.electricalchemy.net/2009/10/cracking-passwords-in-cloud.html
9 comments
This isn't "breaking PGP", it's guessing PGP passphrases. If they had guesses a PGP key using a cluster of virtual machines, that would be news.
it's guessing PGP passphrases
Also, if you can brute-force a PGP passphrase using a cluster of EC2 instances, it's trivial to brute-force it in hardware (using FPGAs or, ideally, ASICs).
If you want to protect something with a passphrase, use scrypt!
Also, if you can brute-force a PGP passphrase using a cluster of EC2 instances, it's trivial to brute-force it in hardware (using FPGAs or, ideally, ASICs).
If you want to protect something with a passphrase, use scrypt!
No matter how you hash them, passphrases are still pretty secure. If you can type your PGP passphrase very quickly, you're doing it wrong.
passphrases are still pretty secure
... at least until someone comes along with a microphone (or a cell phone) and exploits the acoustic side channel. :-)
... at least until someone comes along with a microphone (or a cell phone) and exploits the acoustic side channel. :-)
Hear, hear! They didn't crack or break PGP at all. They still needed the password-protected secret key to pull it off. It's not much different than attempting to brute-force anything else, including login:password pairs.
The passphrase is the weak point in PGP. I suppose some people treat it like a password, which is the wrong approach. It should be much longer and harder than a normal password.
Their analysis does demonstrate how cost-effective it has become to crack low-complexity passwords.
Their analysis does demonstrate how cost-effective it has become to crack low-complexity passwords.
They would also need a way to validate the data unless the encryption was symmetric, I don't think they have tested this out on data they didn't already have.
They were brute-forcing a PGP-encrypted ZIP archive, which would have been symmetric encryption.
These kinds of archives were produced by the PGP Desktop application (at least in earlier versions; I don't know if it still does) and were functionally similar to using "gpg -c", although the actual file format is/was different and incompatible.
Had the guy used a public/private keypair for encryption rather than a password, I suspect the work needed to bruteforce it would be much greater. But that doesn't look like the situation.
These kinds of archives were produced by the PGP Desktop application (at least in earlier versions; I don't know if it still does) and were functionally similar to using "gpg -c", although the actual file format is/was different and incompatible.
Had the guy used a public/private keypair for encryption rather than a password, I suspect the work needed to bruteforce it would be much greater. But that doesn't look like the situation.