Hacker News Reader

Comment by zanny | Hacker News Reader

I was under the impression that even a 128 bit keyspace is prohibitively hard to crack using modern hardware, and that 256 bit would take more computational resources than we can reasonably expect to have available to us for the rest of time. Isn't 1024 bit a bit overkill?

View on HN

RSA can be cracked by breaking the public key down into it's prime factors to discover the private key. AES is much more expensive per "try" because the algorithm itself includes many more steps to do a decryption even with the correct key.

Not the most detailed answer I grant you, but I don't understand all of the math involved.

When you see "256 bit" used in the context of SSL, this is referring to the block cypher used in the actual transmission rather than the RSA key size.

So an SSL implementation will probably use a 2048 bit RSA key and a 256 bit AES key. RSA is used to swap the shared secret which is used as the key to the main AES tranfer.

Thanks, I forgot asymmetric keys were easily computed. I'm just thinking random number space brute force against something like SHA2 SSL.

For AES you have to search the entire keyspace, but for RSA ( and AFAIK any public key scheme) you can actually calculate the private key. The security of public key cryptography relies therefore on one-way functions, which are easy to calculate in one direction but very hard to calculate in the other direction. ( This is used to generate the public and private keys together, but the inverse operation, breaking the encryption, should be prohibitively hard to calculate in practice.) Because of this you can not really compare key length between symmetric and asymmetric encryption schemes.

You're confusing symmetric key sizes with asymmetric key sizes. 1024 bits is a huge symmetric key, but a rather small asymmetric key.

The 128 bit strong key is applicable for symmetric key encryption (e.g. AES, Blowfish etc.). However, even 2048 bit keys are considered relatively insecure for asymmetric key encryption (e.g. RSA).

Reply on news.ycombinator.com