java - Faster asymmetric cipher than RSA -
I am working on a system that uses asymmetric encryption in a large number of files. I am using RSA with 4096-bit keys to encrypt a 256-bit randomly-generated AES key, but there is a lack of performance, because an operation required for all files (estimated number of scans during scanning To do ) Use is about 10,000) and identify how can be decrypted using a specific private key, though I hope this operation will not be immediate, it is taking too long (~ 2 files processed per second) ). I considered reducing the important length, but even taking it to 2048 bits does not provide the level of performance that I need, it will be about 512 bits to be deducted, but as the keys are still trivial Which are out of the question.
Does anyone tell me in the direction of the system that is fast, but the same cryptographic strength? This will need to be implemented by Java JCA provider (like some Bouncicastel) to plug my existing application neatly. I know that the bouncy palace supports El Gammal, but I am not getting any information about this algorithm, or if it is likely to be faster than RSA. I also hear about elliptic curve systems which only require relatively small keys (384 bit or similar), but none of these know where to implement.
For the FAQ, Diffie-Hellman is also known as "ECDH" on elliptic curves. is. Once we want to deal with those shapes, which can not be broken by the current technique, the security estimation is a bit difficult because it depends on how we can bet on the future of technological development. Still, one can say that ECDH provides "128 bit" of protection on the P-256 curve, which is similar to that of a level 2048-bit RSA. This level is widely enough for all current uses, or more reasonably, if P-256 is not enough for you, then your problem is very special and the cryptographic strength is your least concern. There is a possibility.
On its PC (2.4GHz Intel Core 2, 64-bit mode, Linux is running), OpenSSL claims a reduction of 900 ECDH examples per second, using a single core. Edit
Comments
Post a Comment