The preceding defense prevents attackers from disrupting a large portion of our system; it does nothing, however, to prevent attackers from targeting one or two hated machines. It is not clear that much can be done to defend against such a focused attack. By requiring that all a computer's partners be located on different IP subnets, forcing partners to be chosen randomly from the eligible subset, and charging a fee for (excessive) switching the cost of such an attack could be raised somewhat, but probably not enough to deter determined attackers.
The attack is unlikely to be used, however, because it is not very effective at damaging the target machine compared to alternative attacks such as targeted viruses and denial-of-service attacks: it only blocks restoration; actual data loss requires an independent event that might take years to happen, if it happens at all. There is little point in combining this attack with a method for destroying a computer's data because it is usually easier to modify such methods to destroy any backup directly by corrupting the backup software.