Solve network problems with multiple routing protocols

For various reasons, many networks adopt multi-route protocols. This phenomenon is usually caused by merger and acquisition activities, or temporary occurrence during the migration.

Many vro hardware manufacturers may use a variety of proprietary protocols to connect with standard protocols. Some routers will run more than one protocol and inevitably receive the same route from the two protocols. They must make a decision between the two routes when forwarding this packet. The potential problem here is that these protocols may have different "next hops", which can easily lead to routing loops or black hole communication problems. When you make changes and everything works properly, this kind of problem is especially frustrating. However, some modifications to the other part of the Network will generate a cascade effect on one of the protocols and change its next hop. This fault is very difficult to solve, because it is likely that this "problem" may appear hundreds of miles away from the place where you modified it, and that region is not included in the test plan you modified.

Readers who frequently read the content on the SearchNetworking.com website will no doubt know about Cisco routers. This decision is made using Administrative Distance. Here, the Protocol is assigned an 8-bit value. The lower the value, the more "trusted" the protocol is ". For example, a route received through the External Border Gateway Protocol (E-BGP) (AD = 20) takes precedence over OSPF (Open Shortest Path Protocol) (AD = 110) the received route is selected.

However, when designing a multi-protocol environment and troubleshooting faults in a multi-protocol environment, especially when planning specific migration steps, it is often easy to forget that the management distance is not the first decision in this process.

For example, if you plan to transfer your network from one protocol to another, the most important thing for you is to realize that the length of the prefix is more important than the management distance. For example, consider the following route:

192.168.1.0/24 via RIP (AD = 120)

192.168.1.0/20 via BGP (AD = 20)

In this example, your router will send this packet to the next hop recognized by the RIP route, rather than the next hop recognized by the BGP Route, even though the bgp ad is 20. This is because long prefixes are always prioritized. AD takes effect only when the route length is the same.

Why is this important? This is the most common trap when you are transferring from a classless protocol to a classless protocol in a subnet. It is common to receive routes with different lengths. Of course, as long as you realize what will happen and are ready for it, there is no problem.

The policy you can take to solve the problem is to make the routing length completely consistent by summarizing the smaller routes. Obviously, this is not easy in every situation, but it is a choice. Another option is to make more obvious boundaries between your route selection domains. You are not allowed to run two Protocols at the same time on all vrouters in the network in these routing selection domains. You are only allowed to run the protocols that connect these routes to the selected domain. Make sure to carefully filter the broadcast between the routing selection domains.