nilloc (678273) writes "Anyone using the NEW AT&T as their internet service provider was down Monday night in the Southeastern part of the United States of American. I thought the primary design feature criteria for TCP/IP was that the traffic is routable. Therefore if a nuclear attack occurred that took out part of the pipe, traffic would just get re-routed. So I am asking the Slashdot community, why is AT&T (a major Internet Service Provider) putting all of their DNS servers behind one faulty router? Where does the redundancy lies?"