Ideas for improved effort...

Adam Haberlach (
Sun, 30 Mar 1997 19:53:06 -0800

It seems like this effort is really getting off of the ground.
We're 1/10 of a percent through the keyspace (let's have a party at
1%!). We seem to be getting attention via word of mouth, and most of
all, we have a lot of different organizations beginning to show
interest. Once we have a few people within a given organization, it is
usually fairly easy to pass word around and get more machines running.
I was the first person here at .orst to start running the client
(on my laptop at home) after reading about it on one of the crypt lists
that I'm subscribed to. So far we've managed to get the lab I work for
to contribute significant cycles to it, and the college of business let
us use their machines for spring break--all we needed to do was ask.
We're working on getting the Computer Science and Engineering
departments behind us as we speak. Unfortunately, abour 100 pentiums go
back to running login screens tomorrow morning, so I expect that our
keys/day will fall considerably.

One of the things that made it so easy was the simplicity of the
client--just run it and go. We ended up putting it into login scripts
for most of the computers, making it trivial to fire up 100+ machines.
We might be able to do this every evening when the lab closes.

However, it would be nice to have more client options. It has
been said that clients should be able to check out a larger set of keys
(perhaps a day's worth) at a time, or use other methods of accessing the
keyserver. Right now, network traffic at this end is negligable, but as
we get more and more machines sending a request every 30 minutes, we're
going to hit the keyserver harder. Has anyone figured out how many
clients it can handle? How is it holding up? It might be worth it to
make the auto-ranged keysize adjustable so that our steady clients don't
hit it so often.

Do we have a bi-directional protocol of any kind of the server?
If we could have the client query the server for progress stats, we
could write a cool screensaver that people would sit and watch while it
calculated keys. Even if we can't query the server, it would be a good
way to make it easy for people to run the client--just tell them to
install a screensaver, and every time they look away from the computer
for 30 minutes, we get more keys checked. Does someone want to code
this? Other options would be to have a Windows NT service, so it could
run under an NT login screen.

Also, has anyone been tracking any of the other efforts? Are we
ahead? Are we gaining? Maybe we can get some media attention for being
the first to hit 1% or 5% of the keyspace. If we could get a blurb in or one of the other services, we could probably
get some major hits from the curious and cyber-savvy willing to fire up
the client every once in a while. Or even once. 1000 people firing it
up for 60 minutes is about four machine-days of work.

Adam Haberlach
Crack DES NOW!