Re: [tlug] eradicate the evil with quality entropy

[Bruce Howard <bruce@example.com>]

On Dec 22, 2005, at 12:04 PM, Jim wrote:

> Edward Middleton wrote:
>
>> Unless you have a really good entropy source I would suggest urandom.
>
> It's a classic speed versus quality issue. Choose your poison.
>
> Since one should wait about three weeks for each pattern to soak in,
> the slowness of /dev/random should not be a problem.
>
> You can increase entropy inputs with audio noise in some audio input.
> Wind and empty spot on AM radio work. Put one's network card in
> promiscuous mode on a busy network that uses a hub instead of a switch.
> Even with a good entropy source, /dev/random is slower than 
> /dev/urandom.
>
> If you want speed, use a grinder.

I've wondered about this.  Is this three week soak thing really an 
issue at current media densities?  We pack so much information per unit 
of space on the platter these days it seems like a few rewrites ought 
to be sufficient to blow away and remnants of previous data.

A colleague of mine in the United States recently claimed that 
commercial data recovery services will no longer even accept drives 
above a certain density as it is simply to hard to recover anything 
useful.

									Cheers,
									Bruce