In-depth explanation of slow convergence of the RIP Routing Protocol

The use of RIP is the most common protocol. This protocol is also the root cause of many routing protocols. To better understand the routing knowledge, we are now imagining a network topology based on the RIP routing protocol. If R1 is directly connected to network 1, there is a route to the network with a distance of 1 in its routing table. This route is included in the periodic routing broadcast. r2 learns the route from R1 and sets up a route operator in its routing table to broadcast the route value 2. finally, R3 learns the route from R2 and broadcasts it with a distance value of 3.

Now we assume that the connection from R1 to network 1 is invalid. then R1 immediately updates its routing table to set the routing distance to 16 infinity ). during the next broadcast, R1 should announce this information. however, unless the Protocol includes additional mechanisms to prevent such situations, other routers may broadcast their routes before R1 broadcast. it may be assumed that R2 advertised the route exactly after the connection between R1 and network 1 fails. therefore, R1 receives the packets from R2 and uses the common vector distance algorithm: it notices that R2 has a lower cost route to reach network 1, calculate the cost of a 3-hop R2 announcement on the arrival of network 1 is 2 hops, plus 1 hop on the R2 ). then, load a new route that reaches network 1 through R2 in the routing table. in this way, after receiving the datagram from network 1, any one of R1 and R2 will transmit the message back and forth between the two until the lifetime timer times out and overflows.

The RIP routing protocol subsequently broadcast by the two routers cannot solve this problem quickly. during the next round of traffic routing information exchange, R1 announces the various items in its routing table. after R2 learns that the distance from R1 to network 1 is 3, it calculates the new length of the route 4. in the third round, R1 receives the information about the route distance increased from R2 and increases the distance of the route in its route table to 5. until the distance reaches the RIP limit.

Slow convergence

Split horizon update can be used to solve the slow convergence problem. when the range separation technology is used, the router records the interfaces that receive the routes. When the router advertises a route, it does not send the route back through that interface. in this example, vror2 R2 does not advertise the route from 1 to 2 to R1. Therefore, once the connection between R1 and network 1 fails, it will no longer advertise the route. after several rounds of Route Selection updates, all machines will know that network 1 is inaccessible. however, the split range update technology cannot solve all topology problems.

Another way to consider slow convergence is to use the concept of information flow. if the router advertises a short circuit route to a network, all the receiving routers will immediately install the route. when a router stops notifying a route, the protocol should work according to the timeout mechanism before determining that the route is not reachable. when a timeout occurs, the router looks for an alternative route and starts to spread this information. unfortunately, the router does not know whether the alternative route depends on the route that just disappears. therefore, we should not spread the negative information quickly. there is a warning sentence or a sentence:

Good news spread fast, and bad news spread slowly.

Another technique that solves the slow convergence problem uses the hold down method. the suppression method forces the router involved in the protocol to ignore any route information about the network within a fixed period of time after receiving the information about the network's inaccessibility. the typical suppression duration is 60 seconds. the idea of this technology is to wait for enough time to make sure that all machines receive bad messages and will not mistakenly accept outdated messages. it should be noted that all the machines involved in the RIP routing protocol must follow the suppression policy, otherwise the routing loop will still occur. the disadvantage of the suppression technology is that if a routing loop is found, the routing loop will be maintained during the suppression period. more seriously, all the incorrect routes are retained during the suppression period, even if there is an alternative route.

The last technique to solve the slow convergence problem is the toxic reversal of poison reverse ). when a connection disappears, the router retains the route for several update cycles. However, when a route is broadcast, the cost of the route is set to be infinite. to improve the efficiency of the toxicity reversal method, It should be combined with triggered updates. the update trigger technology reduces the time for errors due to messages of good news.

Unfortunately, although trigger update technology, toxicity reversal technology, inhibition technology, and split-range technology can solve some problems, they bring about some new problems. for example, when many routers share a public network structure and adopt the update trigger technology, a broadcast can change the router routing table and lead to a new broadcast. if the second round of broadcast changes the route table, it will cause more broadcasts. this produces a broadcast avalanche.

The use of broadcast technology may lead to a routing loop) and the use of suppression technology to prevent slow convergence problems, which can make the efficiency of RIP on the Wide Area Network very low. broadcasting consumes a lot of valuable bandwidth. even if no broadcast avalanche occurs, periodic broadcast of all machines means that network traffic increases as the number of routers increases. the possible routing loop may be fatal when the line capacity is limited. when the packets in circles make the capacity of the line saturated, it is very difficult or even impossible for the router to exchange some routing packets to break this loop. similarly, in the wide area network, the suppression period may be too long, so that the timer used by the high-level protocol times out and thus the connection is interrupted. despite these well-known problems, many organizations still use RIP as IGP on the wide area network.