In a previous thread I discovered that GPG performance for encryption with public/private keys is similar to the performance for symmetric encryption. GPG generates a random passphrase (random number of bits matching the symmetric cipher), encrypts that passphrase with the private key, and then encrypts the file with the passphrase.
Signatures, on the other hand, will always be affected by the key size so. But I don't know what kind of signature generation is performed by duplicity.
Anyway, you can always delete the cache and perform a local test (local target directory) to check the differences.
Hi,
I'm wondering whether a GPG key size of 2048 or 4096 bit will make a difference on the encryption/signing or decryption/verifying performance in duplicity. I'm used to much shorter backup and restore times without using an asymmetric key in duplicity, i.e. only a password. Then I started with a 4096 bit key and added signing and things got much slower, especially for the restore process (4 to 30 minutes).
So I redid it all and used a 2048 bit key. But today as I try that, it's even a lot slower. I believe that remote storage (SFTP) performance may have a greater effect on this, so my test method won't work.
Maybe SFTP is also slower to transfer than FTP which I used before, without the GPG key.
Does anybody have experience with key size vs. backup/restore speed? In general, if I can get higher security, I'd like to use it. But if it makes a full restore take double the time, I might reconsider this.
--
Yves Goergen
http://unclassified.de
http://dev.unclassified.de
______________________________
_________________
Duplicity-talk mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/duplicity-talk