An American expert invented a "Digital Cannon" to destroy the entire internet.

The British New Scientist Weekly reported on February 11 the original question: Network weapons that can destroy the Internet (reporter Jacob Allen)
A new network weapon can destroy the entire Internet-and there is currently almost no defense to block it. He spoke to his colleagues at the University of Minnesota about the digital cannon they created ". Fortunately, they do not intend to destroy the internet. Instead, they are proposing to improve Internet defense.

Internet structure storage Vulnerability

Shuhad's new attack method uses the structure of the Internet to attack itself. On the network, many nodes are offline every minute, but we will not notice that the network will bypass them. It can do this because the smaller networks that make up the Internet, known as autonomous systems, communicate with each other through routers. When a communication route changes, nearby routers send notifications to nearby routers through a so-called Border Gateway Protocol (BGP) system. These routers then send notifications to other neighboring routers, and finally release the new path to the entire internet.

A previously discovered attack method called the ZMW attack is to disrupt BGP, so that the connection between the two routers is displayed offline, thus cutting off the connection between the two routers. Shu had and his colleagues found out how to extend this method to the Internet and simulate its effect.

This attack requires a huge "botnet"-a network composed of computers infected with Trojans. Shuhad estimated that 0.25 million computers would be enough to destroy the internet. Botnets are often used to launch distributed denial-of-service (DDoS) attacks, which cause them to crash by overloading network server traffic. However, this new attack method of shuhad is different.

"Digital Cannon" Operating Mechanism

Attackers who launch the shuhad network weapon need to send traffic between computers in the botnet and create a "path map" between them ". Then they need to find a connection shared by many paths and launch the zm w attack to destroy it. The nearby router will respond to this and send

BGP updates messages to direct traffic to other places. After a short period of time, the two disconnected routers will reconnect and send their own bgp update information. The attack traffic will flow in again to disconnect them again. This cycle repeats, and each disconnection and re-connection will send bgp update messages to each vro on the Internet. At last, each vro in the world receives an update message that exceeds its processing capability.

When every vro in the world is occupied, normal route interruptions cannot be repaired. In the end, the Internet will become riddled with holes and cannot communicate with each other. He believes that this situation takes several days to recover.

"Once such an attack is launched, it cannot be solved through technical means. network operators can only communicate with each other orally," he said ." Each autonomous system must be shut down and restarted to clear the BGP backlog.

How to Prevent Network crashes

Is the Internet crash inevitable? Probably not. This type of attack is unlikely to be initiated by hackers, because it is a highly technical task to draw network maps and find target connections, in addition, anyone with enough botnets is more likely to rent it out for profit.

No matter who launches such an attack, we can't do anything about it. Shu had's simulation shows that the existing BGP built-in fault protection measures are almost powerless for his attacks. One solution is to send bgp update messages through an independent network, but this is not realistic because it is necessary to establish a shadow Internet.

Another way is to change the BGP system so that it can assume that the connection will never be disconnected. However, according to the investigator's model, this method must make at least 10% of autonomous systems on the Internet, network Operators are also required to find other methods to monitor the health status of the connection. He said it would be difficult to persuade enough independent operators to make the change.