[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-talk] Tor Survey

The posted script uses the "openssl speed" command to evaluate the node's encryption performance. This command iterates through every algorithm OpenSSL supports, most of which Tor doesn't use. There is also a bug in OpenSSL where the speed command alone does not utilize encryption accelerators, so the output of this part of the script is completely useless to the study anyway. To get the actual performance Tor sees you have to test each algorithm Tor uses independently with "-evp". For example, on one of my Tor nodes "openssl speed" gives me:

type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes aes-256 cbc 66979.82k 72147.67k 75580.48k 72884.57k 72832.34k

while "openssl speed -evp aes-256-cbc" gives:

aes-256-cbc 442163.07k 472286.16k 477768.11k 485609.13k 484977.32k

So Tor actually has available to it over 6x the performance for this algorithm of what the script will show. No idea which algorithms Tor actually uses though.


On 1/11/2012 4:47 AM, Marco Valerio Barbera wrote:
Dear Tor Relay Administrator,

my name is Marco Valerio Barbera, I am a PhD student in Computer Science at La Sapienza University of Rome and I am currently doing a research study on the security of the Tor Network in collaboration with Angelos Keromytis, head of the Network Security Lab at Columbia University.
--- snip ---
tor-talk mailing list