ietf-openpgp
[Top] [All Lists]

Re: Reasons to include ECC to our charter

2001-09-04 15:41:01

At 5:56 PM -0400 9/4/01, vedaal wrote:

but that would be a very noticeable increase in speed  of encryption of
very large files,
or pgp disk encryption at the gb level of pgp disk size

agree fully with the principle of including it as an option even though it
is patented.

once the advantages become clear enough, someone will hopefully find a way
to produce a funtionally similar patentless alternative


ECC does *not* affect the speed of large file encryption. Actually, it's
the opposite; it matters most for *small* files.

Remember, the bulk encryption is done with a block cipher. If you want it
to be fast, you use CAST, Twofish, or Rijndael. All of them are very fast.
You might spend 50ms with the public key packet, and then 1 or 2 ms with
the data.

The symmetric key is then wrapped in the public key encryption. For a small
file, you spend a lot of time with the wrapper and then pop open the data
packet. With a big file, the heavy cost of the public key operation is a
smaller fraction of the total cost, and ECC matters less. ECC is a win when
you have teeny processors sending small messages.

        Jon