Route summarization in Cisco Networking

As already mentioned, classful routing protocols automatically summarized networks to the classful subnet boundaries. Classless routing protocols, on the other hand, require you to manually control the networks being summarized to your neighbors in the router configuration. By aggregating a contiguous set of networks into an advertised summarized route, you keep the size of the routing tables to a minimum.

Neighbors that receive the summarized route do not need to know about the individual subnets you create behind your router because they inevitably have to go through your router to get to them. The additional offshoot of this summarized picture is that your classless routing protocols do not need to notify those neighbors if one of those subnets goes down because they do not even have that subnet in their routing tables. Thus, you can isolate topology changes to be contained behind that summarizing router.

Because you are required to manually specify the networks you are to advertise, you must learn how to accurately summarize smaller subnets into one or several larger networks, or supernets. The rules for supernetting are similar to subnetting, except, in this case, you are stealing bits from the network portion of an IP network to create a larger network. The rules for super netting are as follows:

1. Be sure that the networks are contiguous (otherwise you would be summarizing networks that you do not have behind the router.)
2. Count the number of networks you want to summarize.
3. Determine an increment that is equal to or less than the number of networks.
4. Make sure your base networks start on incremental boundaries (128, 64, 32, and so on) for the number of networks you are summarizing.
5. Calculate the subnet mask by the number of bits you need to steal from the original subnet to equal that incremental value.

The beauty of supernetting is that the resultant network and subnet mask will designate many IP address networks in a single entry. The fact that you are stealing bits from the network portion of an IP address could quite easily violate the traditional barriers of classful addressing, known as Classless Interdomain Routing (CIDR).

For instance, it would not be uncommon to see a summary entry look like the following: 192.168.160.0/20. This single entry used to be a class C (/24), but four bits were stolen from the network portion to represent 16 networks (24=16). When you advertise this supernet to neighbors, they know that they must go through your router to get to networks 192.168.16.0 through 192.168.31.0(16 total networks).

The following figure illustrates a typical route summarization example in which Router B is summarizing all its subnetted networks to Router A as one supernetted network. Following the steps outlined previously, you can determine the aggregate network entry to advertise, as follows:

• The networks are all contiguous, so you can summarize them accurately.
• A total of 32 networks need to be summarized.
• 32 conveniently falls on an incremental boundary.
• Because the network if 192.168.64.0, 64 is an increment of 32 so we can use that as the base network for the summary route.
• You must steal 5 bits (25=32) from the /24 network, so /19 (24 5=19).

By creating the summary route 192.168.64.0/19, Router A is required to maintain only that one entry in its routing table as opposed to the individual 32 subnets. If a topology change occurs in one of the subnets behind Router B, there is no need to advertise that changes to Router A because it knows about only the summarized network.

Interior and Exterior Gateway Routing Protocols

Routing protocols can fall under two major categories depending on the autonomy of the network on which the routing protocol exists. The identifying characteristics of the category to which the routing protocol belongs ultimately depends on whether the routing protocol exchanges updates within a network that is under your administrative control. When the network is under your control in your administrative domain, it is known as an autonomous system.

Routing protocols used to disseminate information to maintain routing tables and establish pathways inside an autonomous system categorized as Interior Gateway Protocols (IGPs). Conversely, the other category of routing protocols is designed to route in between these autonomous systems.

For instance, Broder Gateway Protocol (BGP) is a routing protocol that is used by ISPs for routing traffic over the Internet. Because the Internet comprises thousands of networks, each under different administrative control, you need to use an Exterior Gateway Protocol such as BGP to route in between these autonomous systems.

Distance vector routing protocols

In addition to being an IGP/EGP or classful or classless, routing protocols can also fall into one of three classes. Again, the functionality and characteristics of the routing protocol dictate under which class it falls. The most long-standing of these classes are distance vector routing protocols.

Distance vector routing protocols concern themselves with the direction (vector) in which the destination lies and some means of measurement (metric) it takes to reach that destination. Distance vector routing protocols informs their directly connected neighbors of all the connected and learned networks they know about in their routing tables.

In fact, they broadcast the contents of the entire routing table to their neighbors periodically, regardless of whether there is a change in the network topology. When the neighbors receive that routing information, they identify and add any new networks to their routing tables and update the metric before eventually passing it on to their neighbors.

Because the routing table information is updated before it is sent on to neighbors, downstream routers do not learn that information first hand. For this reason, distance vector routing protocol update processing is often referred to as “routing by rumor”.

Link-state routing protocols

As the name states, link-state routing protocols advertise the state of the links in the network. In fact, they advertise the states and metrics (cost) of all the links they know about for the entire topology to their neighbors, as opposed to just the best routes in your routing table.

This detailed overview of the entire routing domain enables each router to calculate and make a decision on the best route from this first-hand information, rather than listen to what its neighbors believe is the best route.

In fact, link-state routing protocols keep three tables; a neighbor table of all discovered neighbors, a topology table of all the possible routes to reachable networks learned, and a routing table that contains the best route based on the lowest metric calculated from the topology table.

At first, this may sound like a lot of information to be exchanged between routers; however, link-state routing protocols initially discover their neighbors when they first boot up and synchronize their topology tables.

After the neighbor discovery and topology synchronization, they send only periodic hello messages to let their neighbors know they are still functioning. This is significantly different from distance vector routing protocols that periodically exchange the entire routing table, which can contain a large amount of information, depending on the size of the network.

In addition, link-state routing protocols react much faster when a topology change occurs in the network. In fact, these protocols were initially created in response to the slow convergence issues that you typically encounter with distance vector routing protocols. The downfall of these routing protocols is the resources they consume in the router. Namely, maintaining and processing three tables consume quite a bit of memory and processor power.

Advanced distance vector/Hybrid routing protocols

they say it usually takes three tries to get something absolutely right. The truth behind this saying is that you learn from the mistakes of the previous two attempts. Such is the case with advanced distance-vector, often referred to as hybrid or balanced hybrid routing protocols. Because they take the best features and avoid and pitfalls of both distance vector and link-state routing protocols, hybrid routing protocols are a more proficient breed of routing protocols than their predecessors.

The Routing Table Revisited

Now that you have learned about several types of routing sources, including Static routes and Dynamic Routes protocols, it’s time to revisit the routing table and solidify how network entries are added and used in routing decisions. To help illustrates this process, refer to the following show IP route output:

Router A>show ip route
Codes: C-connected, S-static, I-IGRP, R-RIP, M-Mobile, B-BGP, D-EIGRP, EX-EIGRP external, O-OSPF, IA-OSPF inter area, N1-OSPF NSSA external type 1, N2-OSPF NSSA external type 2
E1-OSPF external type 1, E2-OSPF external type 2, E-EGP, i-IS-LS, L1-IS-IS level-1, L2-IS-IS level-2, *-Candidate default U-per-user static route, o-ODR, P-Periodic downloaded static route T- traffic engineered route.

Gateway of last resort is not set

R 172.17.0.0/16 [120/1] via 192.168.1.10, serial0/0
C 172.16.0.0/16 is directly connected, FastEthernet0/0
10.0.0.0/18 is variably subnetted, 4 subnets, 3 masks
D 10.2.0.0/16 [90/2297856] via 192.168.1.10, serial0/0/0
D 10.3.0.0/16 [90/2297856] via 192.168.1.10, serial0/0/0
R 10.0.0.0/8 [120/2] via 192.168.1.6, serial0/0/1
D 10.1.0.0/16 [90/2297856] via 192.168.1.10, serial0/0/0
D 10.4.0.0/16 [90/2297856] via 192.168.1.10, serial0/0/0
O 10.4.0.0/32 [110/65] via 172.16.0.1, FastEthernet0/0
192.168.1.0/30 is subnetted, 2 subnets
C 192.168.1.8 is directly connected, Serial0/0/0
C 192.168.1.4 is directly connected, Serial0/0/1

Notice that there are now several entries for directly connected networks, a static route, and several dynamic routing protocol entries from EIGRP, RIP and OSPF. For each dynamic routing protocol, the network and subnet mask are being advertised by neighbor routers, followed by two numbers in brackets separated by a slash (/).

The number to the left of the forward slash is the administrative distance of the routing protocol. The number to the right of the forward slash represents the metric that is being used by the routing protocol to determine the best path to the destination network. This information is immediately followed by the router from which it learned this information (thus, the next-hop address). The last item in the routing entry represents the interface packets that must exit to reach those networks.

Assuming that several of the routing protocols advertised the same networks, how did these specific network entries come to be in the routing table? The obvious answer is that the interfaces, a static route, and multiple routing protocols were configured and the resultant table just appeared. However, to answer the question more specifically, each routing protocol determined which routes should be entered in the routing table based on the lowest matric to those destinations. In the chance that one or more routing sources is trying to place a network entry in the routing table for exactly the same subnet, the routing protocol with the lowest administrative distance is chosen because it is the most trustworthy.

After the routing table is built, packets are routed to their destinations by examination of the destination IP address in an IP packet and associating the network in the routing table with that IP address.

If there isn’t a match for the network lookup, the packet is forwarded to its default route. If the gateway of last resort is not set (as in this show IP route output), the packet is dropped, and an ICMP destination unreachable message is sent back to the source to indicate that the destination cannot be reached. In the show ip route output, several entries for the 10.0.0.0 network are listed in the routing table.

Interestingly, there is a RIP entry for the 10.0.0.0/8 network to go out serial0/0/1 and four EIGRP-learned networks for 10.1.0.0/16, 10.2.0.0/16, 10.3.0.0/16, and 10.4.0.0/16, all destined for interface serial 0/0/0. Because the EIGRP networks are subnets of the major 10.0.0.0 network, which interface will the router use to route a packet destined, for example, for 10.1.0.3?

Cisco’s routing logic answers this question by using a rule called the longest match. The longest match rule states that when a packet has multiple possible network entries to use, the more specific subnet is used over the less specific. In other words, the longer the number of bits in the subnet mask (thus the smaller subnet), the more chance it has of being the chosen network. In the routing table example, a packet destined for 10.1.0.3 would use the subnet with the longest prefix (subnet mask), which is the EIGRP route for 10.1.0.0/16 existing interface serial 0/0/0.

Routing redistribution

You are likely to encounter in your Cisco travels certain situations in which you must run multiple routing protocols in your network. For instance, your company is in the process of merging with another company’s network, and their routers are running a different routing protocol than yours. In addition, you may have to connect your Cisco router network to a non-Cisco routing infrastructure and you are using Cisco proprietary routing protocols.

In instances where you are running multiple routing protocols, it may be necessary to have networks advertised in one routing protocol injected into the other. Unfortunately, because routing protocols are so diverse in nature, they do not inherently interact or exchange information with each other when multiple routing protocols are running in the network. The transferal of network information from one routing protocol into another is a manual configuration called redistribution.

The redistribution configuration is typically done at one or a couple of routers that sit on the boundary between each routing protocol, as illustrated in the following figure. These devices run both routing protocol into the next. Redistribution can occur in one of the two fashions.

• One-way redistribution: networks from an edge protocol are injected into a more robust core routing protocol, but not the other way around. This method is that fastest way to perform redistribution.
• Two way redistribution: networks from each routing protocol are injected into the other. This is the least preferred mehod because it is possible that sub optimal routing or routing loops may occur because of the network design or the difference in convergence times when a topology change to occurs. The following figure which is an example of two way distribution.