Someone said that in today’s network communication traffic, about 80% of the resources are wasted, and less than one fifth of the resources are effectively used. One of the culprits of this resource waste is the generation of network loops, which not only waste the CPU and memory resources of the device, but also the bandwidth resources of the link.
The anti-ring mechanism is the purification mechanism produced in such a bad environment. Generally, our network loops are divided into layer 2 loops and layer 3 loops. The formation of all loops is caused by confusion caused by unclear destination paths.
The Layer 2 loop is also known as the switch loop. It is generated by the vicious circle of broadcast messages of the switch. The layer 3 loop is the loop generated on the router.
The distance vector routing protocol simply announces its own routing table periodically, loads the received effective routes into the routing table, and reflects the distance to the target network through the accumulated metrics. Therefore, the router running the distance vector routing protocol does not know the topology of the entire network. These characteristics make it very easy to have routing loops in the network.
When the X network on the R1 router side fails, the R1 router receives the failure information, sets the X network as unreachable, and waits for the update cycle to notify the adjacent R2 routers. However, if the update cycle of the adjacent R2 router comes first, the R1 router will learn the route to the X network from the R2 router, which is called the wrong route, because the X network has been damaged at this time, while the R1 router has added a route to the X network through the R2 router in its own routing table. Then the R1 router will continue to notify the R2 router of the wrong route, and the R2 router updates the routing table, thinking that the R1 route must be passed to reach the X network, and then continues to notify adjacent routers, so that the routing loop is formed.
To sum up, the problem of routing loop is very harmful to the network. Therefore, from the perspective of network design and protocol design, we should fully consider the hidden dangers and possibilities of the loop and avoid them.
You can also read clever dpscd
Solution to the layer 3 loop:
1. Define the maximum number of hops:
In order to avoid the endless forwarding of RIP routes in the network, the maximum hops of a route are defined as 15 hops. That is, when the metric value of a route reaches 16 hops, the route is considered unavailable and the network segment to which the route points is considered unreachable.
2. Horizontal split:
A method to eliminate routing loops and speed up network convergence is implemented through a technology called “horizontal segmentation”. The rule is not to send the route update information again in the direction of the original route update.
3. Route poisoning:
When a network becomes unreachable, the router that finds the change immediately triggers a 16 hop route update to notify the routers in the network that the target network is unreachable. This route is called toxic routing.
4. Toxicity reversal:
After learning routes from an interface, it will carry these routes when it sends a response message from the interface, but these route metrics are set to 16 hops. You can clear useless routes in the opposite routing table.
5. Trigger Update:
Normally, the router will send the routing table to the neighbor router regularly. Triggering the update is to immediately send the route update information in response to some changes. The router that detects a network failure will immediately send an update message to the neighbor routers. And then generate the neighbor routers that trigger the update notification in turn, so that the routers on the entire network can receive the update message in the shortest time, so that they can quickly understand the changes of the entire network.
