This is a good question that pertains to any offline 
site back-up tool. If you want to stop HTTrack from 
overloading your server from a SINGLE connection (ie, 
one T3 line) you should probably just follow Xavier's 
advice on server configuration: just limit bandwidth 
per IP. but: there is a program called Multiproxy that 
can be hooked up through HTTrack, allowing HTTrack to 
connect through many multiple proxies, and making it 
near impossible to block each incomming IP. but, there 
is some server-side pattern recognition software that 
should help allievate this problem by watching for 
spiders, but you will have to find such a program on 
your own and configure it on your service.  If your 
goal is to only keep HTTrack from stealing copyrighted 
material off of your site (and yes, people do this 
with HTTrack) then you should set up a few of those 
loops (that caused your server to crash) in different 
places, and limit per IP bandwidth and connections.  
The only thing to watch out for is proxies -- these 
can mean certain death for a server.

- Austin B