ONeill Jack wrote: > scanrand is a scanner like nmap, but it seems to be > very fast. So it may be interesting to include it in > Nessus, doesn't it ?
I'd rather use a regular TCP stack for discovering open services because it's known to work reliably. With kqueue/epoll, it should be possible to implement *very* fast scanners. (Just with poll(2), I reach about one million addresses per hour.) Congestion can indeed result in problems for such scanners. A regular TCP stack won't help that much (except for being slower 8-). It's important to connect the scanning machine to a core router and distribute the targets rather well.
