Maurice Janssen wrote on 19-9-2007 8:07: > I fully agree, I've been doing it for years.
In your earlier posts you're trying to point out that stateless nat will not work, now you've been doing it for years. Clearly we have some kind of misunderstanding, but I'm missing it. Never mind :-) As someone pointed out earlier, early forms of nat didn't keep state. If I recall well, state tables were only introduced in the late nineties. Most software still support stateless forwarding (except those cheap consumer grade routers). So why wouldn't it work? Keeping state is a sensible thing to do for client machines behind a nat box, but for servers there's no need. For a high volume time server it seems the state table is in the way. See all those threads about Turk Telecom spikes on this list. I'm fairly convinced that almost all of those problems were caused by overflowing state tables. What I tried to point out in this thread is that stateless nat (on udp123 only, while keeping state on other traffic) might be a solution. I was *not* trying to invent some new form of nat. Jan _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
