Strategypage writes about that attack and says that I am wrong on this (Well, not me personally. I doubt they are responding to my post.):
With regard to a cyberwar, if all the root servers were physically destroyed, the Internet could actually continue to function, as the vast majority of domain names don't change.
So these sites aren't critical enough to bring down the Internet. I am definitely wrong about this hacking attack and those thirteen computers.
But I'm reasonably sure that there are such critical physical choke points. Recently, a cut cable screwed up Internet traffic on Taiwan for a bit. Taiwan is an extreme example as a place at the end of a major connection with little room for rerouting on high-capacity lines, so it is not typical. But the event shows that destruction of physical links can be effective in theory. So my concept as a theory is not bad. It just depends on whether America, too, is vulnerable to such a loss.
And I still recall reading an article in pre-blog days that our Internet infrastructure is different than the original defense-related web. I read that we have concentrated nodes in single buildings for ease of maintenance and environmental control. So nodes that in the past would be scattered are now concentrated. Then, if a node broke down from an attack, traffic would reroute through another node a hundred miles away. Today, if a node breaks down on its own from a simple non-hostile fault, one ten feet away in the same bulding may be the one that takes over. That's no problem in peacetime. But if the building is destroyed, a lot of backups will go down, too.
Of course, my memory could be faulty. Or the situation could have gotten better since I read that article.
Bottom line: I'm wrong about this particular post. But I'm not willing to discount my memory of what I read about the structure of the Internet. I really should devote some time to research this. I'll report back.
