RIP Loop Avoidance overview

Reprinted from Http://www.cnitblog.com/langog/articles/4578.html

The DV algorithm, also called the vector algorithm, is the routing protocol Rip,igrp and so on Algorithm Foundation.
Its objective is to seek:
D (x,y) =min (d (X,k), D (k,y))
The method is:
Based on the routing information of neighbor nodes, compute the best route of their own. "Routing by rumor" said that is the meaning. Therefore, nodes do not have a global view of the network, and the reliability of routing calculations is based on the routing information of the upstream node.
This algorithm is proved to be convergent in finite time. However, the algorithm itself does not consider the recovery problem caused by a node or network failure in the context of routing. So in order to improve the robustness of the algorithm, several techniques have been added in practical use.
Because the nature of DV is based on neighbor information, compute routing, so in order to prevent the information of neighboring information misleading, add "horizontal segmentation" technology. The technique is that "the information sent from this interface to a neighbor does not contain a route learned from the neighbor." To enhance the effect of this prevention loop, you can add a "toxic reversal", which is "the metric=infinit from the neighbor's message to the neighbor from the interface." These two techniques can solve the problem of loops caused by error messages between neighboring nodes.
The above method cannot be resolved for loops caused by error messages over two nodes. The essence of this is that the node in accordance with the path given by the upstream node can not be judged whether its own path, that is, can not monitor the loop exists. Therefore, the DV algorithm cannot fundamentally prevent the loop from happening. In order to effectively use the algorithm, artificial limit the life time of the loop, that is to specify the size of the network. At the same time, it is also thought of some methods to speed up the existence of loops in the case of routing ring breaking technology. "Trigger Update" is a way to speed up the transmission of network information. In the absence of a loop, this method can increase the rate of convergence, in the case of a loop, you can speed up the error message loop, thus speeding up the error route to reach the upper jump limit and break the loop. "Suppression Time" is a short period of time when the interface receives a route that is worse than the route that was received and used from this interface before, and prevents the route from changing. Its essence is to reduce the misleading of erroneous routing information, the idea is that if the current route becomes worse, it is likely to be the wrong route, that is, it is better to trust its good, do not believe its bad. Moreover, the routing is worse, it is likely the result of routing loops, so adding "suppression time" can prevent the continued spread of routing loops and cause node timeout routing upstream of the loop.
In this, the "suppression of time" understanding is the most rare part. The rfc1058 of RIP does not include this regulatory parameter, but Cisco and other manufacturers have this in their products. Also, Cisco sets this time to be a 180s,6 cycle. Considering that the routing timeout is also 180s, the two are equal and can accelerate the loop's broken loop.
For the DV algorithm, there's a saying, "Good news is fast, bad news is slow," according to Andrew S.tanenbaum's computer network, the reason for this can be attributed to the bad routing, in the absence of horizontal segmentation and other means of processing, the message spread throughout the network time than good news slow. I understand that even with this approach, bad news is always slow to pass in an irreversible loop.
Here, finally put forward a thought, how the above techniques are discovered. This requires careful comparison, find common, pay attention to differences.