We have analyzed and summarized the features of some RIP routing protocols. We found that some phenomena are not conducive to the formation and use of networks. Therefore, we propose two measures to improve the performance of the RIP routing protocol.
1. horizontal segmentation
In the "count to infinity" Phenomenon of the RIP routing protocol, the cause is that A and B transmit "spoofing information" to each other. In this case, we will naturally think that if we can remove these "spoofing information", we will not be able to avoid "count to infinity" to a certain extent. horizontal segmentation is such a solution.
"Normal horizontal segmentation" is: if a route information is learned from Port X, the route update message sent from this port will no longer contain this route information.
"Horizontal segmentation with drug reversal" is: if a route information is learned from Port X, the route update message sent from this port will continue to contain this route information, and set metric of this information to 16.
"Normal horizontal segmentation" can avoid sending spoofing information, reduce the size of route update packets, and save network bandwidth; "horizontal segmentation with drug reversal" can quickly eliminate the routing information loop, but it increases the burden of route update. the two methods can be selected based on the actual situation.
2. Trigger update
The above "horizontal segmentation" can eliminate the cycle of spoofing information between the two routers, but when three or more routers are involved, the effect is limited. The following network is investigated:
+---+ +----+ +-----+ /-----\
| | | C +-------| D | -----|| E ||
| A +------| | | +---- | |
+-+-+ +----+ +---+-+ \-----/
| -- |
| -- |
| -- |
+----+ |
| | |
| B +-----------------------
| |
+----+
E is the target network
For the target network, the router routing information is as follows:
A: 3 C
B: 2 D
C: 2 D
D: 1 direct connection
After a fault is interrupted between D and the target network, B and C can correctly obtain the inaccessible information from D. However, as shown in the preceding routing information, although A does not send error messages to C, A may send route information to B before receiving the network inaccessible information, so that B may think that A can reach the target network through, then, "Count to infinity" appears.
The RIP route protocol update is triggered to improve the above situation. The specific implementation measures are as follows: Once the router detects a network change, it sends an update packet as soon as possible or even immediately, and does not wait until the update period ends. as long as the update trigger speed is fast enough, it can greatly prevent the occurrence of "count to infinity", but this phenomenon is still possible.
After an update is triggered, when the network topology changes, a message update trend similar to "Domino" appears in the network, and finally stop the router that has never changed the path to the target network.