Shareaza

Connections are not kept alive 100% of the time. This means that if a connection is UDP based and no data is send, a short connection interruption might not be detected at all, causing the G1 connections to stay alive threw the interruption.

It just happened again, but this time the sole survivor was a G2 hub connection to France.

So, either both G1 sensitivity to packet loss > G2 sensitivity to packet loss AND G2 sensitivity to packet loss > G1 sensitivity to packet loss, a logical impossibility, or else Shareaza is responsible for these fumble events rather than some other aspect of networking.

No, I'm just saying it does affect all protocols differently as all of them are designed differently concerning their connections and connection keep-alive mechanisms...

Nope, it does not. Because of the way these protocols are designed, a disconnection is only detected if Shareaza tries to use the specific link. And such link activity is not a regular occurrence, but it happens quite randomly, depending on the actual need for communication between your computer and the other connected host. E.g. if there is no need to tell the other party something, you don't notice the other party dropping out (because you have no reason to expect a reply from the other party).