Re: Performance thoughts

[Fyodor <fyodor () insecure org>]

On Wed, Jul 05, 2006 at 01:06:30AM +0200, Diman Todorov wrote:
> >
>
> Yes /dev/urandom should perform better but it is not really random.
> I have run into problems with it's short period before.

Are you sure you aren't thinking of rand() or something else?  I don't
think /dev/urandom has a period.  Or at least not one we're likely to
ever encounter in our lifetimes.  It doesn't block on entropy like
/dev/random, but I think the risk is extremely limited even for most
cases where "secure" random numbers are truly needed.

In any case, even /dev/urandom is overkill for Nmap's needs.  It just
needs "somewhat random looking" IPIDs, sequence numbers, port orders,
etc.

Cheers,
-F


_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev