Route-summarization tools allow engineers to advertise one route that replaces several smaller routes, with the new route matching the same range of addresses. Route-summarization decreases the usage of bandwidth, RAM, and CPU.
This article first examines manual route summarization concepts. The concepts rely on math that uses the same principles as subnetting math, and it relies on a good subnetting plan that assigns subnets in anticipation of future attempts to do route summarization.
The second half of the article then shows a systematic way to find out the summary route that is the best summary route to create when configuring summary routes.
Reasons for using route Summarization
Small networks might have only a few dozen routes in their router’s routing tables, but the larger the internetwork, the larger the number of routes. Some enterprises might have tens of thousands of subnets. As a router’s routing table grows, problems can occur. The tables themselves consume memory in a router.
Routing (packet forwarding) requires the router to match a route in the routing table and searching a longer table generally takes more time and more work by the CPU. Routing protocols require more work to process the routes and more bandwidth to advertise the routes. With a large routing table, it takes more time to troubleshoot problems, because the engineers working on the network need to sift through more information.
Route summarization allows network engineers to help overcome some of these scaling problems by replacing many routes for smaller subnets with one route to what looks like a larger subnet. For example, instead of advertising routes for a lot of /24 subnets, such as 172.20.1.0/24, 172.20.2.0/24, 172.20.3.0/24, and so on, the router might simply advertise a route for 172.20.0.0/16, and not advertise all those smaller subnets.
In the next section you will learn how the subnetting plan needs to be considered in the planning phase in order to implement summarization.
Route Summarization and the IPv4 Subnetting Plan
Route summarization combines multiple routes into one route, but for that to work, the original routes must be in the same numeric range. That can happen by accident, but it works much better with planning.
For example, the figure below shows a sample internetwork, with two sets of four subnets that could be summarized . Note that the subnetting plan placed subnets that begin with 192.168 on the left and those that begin with 172.16 on the right, which makes route summarization a little easier.
To see why, focus on the right side for now, and ignore the subnets on the left. The figure shows the conditions before route summarization, for routes learned by R1, for the subnets on the right.
Manual route summarization causes a router to cease advertising some routes, instead advertising
a route that contains a superset of all the addresses. To do so, the router that creates the summary must be configured to know the subnet number and mask to advertise in a new summary route. The routing protocol stops advertising the old smaller routes (called subordinate routes) and now advertises only the summary route.
The following figure continues the above example by showing the effect of a summary route configured on Router R3. That summary route replaces the routes for all four of the subnets on the right. Just to make the math easier, the summary route uses a subnet of 172.16.0.0/16.
By creating the summary route configuration on R3, R1 (and other routers farther into the network)
receive the benefit. R1’s routing table decreases in size. More importantly, R1 can still forward packets to those same original four subnets, out the same Gi 0/1 interface.
The process of find the best Summary Route
To find the best summary route, you can use trial and error, use a subnet calculator, or use any other method you like. The process uses familiar skills: taking a subnet ID/mask and finding the subnet broadcast address. If you can do that math in your head with confidence, gradually you won’t need any calculator. If not, consider going back to review subnetting math.
Here are the steps for finding the best summary route, using decimal math, with some examples to follow:
- Step 1: List all to-be-summarized (subordinate) subnet numbers in decimal, in order, from lowest to highest, along with their matching subnet broadcast addresses.
- Step 2: Note the low and high end of the range of addresses for all combined subnets by noting the numerically lowest subnet ID and numerically highest subnet broadcast address.
- Step 3: Pick a starting point prefix length /P for Step 4, as follows: Pick the shortest prefix length mask of all the subordinate subnets, and then subtract 1.
- Step 4: Calculate a new potential summary subnet/mask, with matching broadcast address, based on the lowest subordinate subnet ID from the original list and the current prefix length.
A: If the calculated range of addresses includes the entire range from Step 2, you have found the best summary route.
B: If not, subtract 1 from the prefix length used in the most recent calculation and
repeat Step 4.
As usual, the steps themselves can be daunting. Here’s the shorter version:
- Pick the lowest subnet ID from the list, keep shortening the shortest prefix-style mask, calculate a new subnet ID based on those details, and see whether the new subnet includes all the addresses in the list of original subnets.
The best way to really understand the process is to see a few examples.
Example: finding the “Best” Summary on Router R3
R3, in the above figures connects to subnets 172.16.0.0/24, 172.16.1.0/24, 172.16.2.0/24 and 172.16.3.0/24. Below you can see the results of the first three steps, applied to the three routes of Router R3.
- Step 1: Relist the subnet IDs (and prefix lengths) and calculate the subnet broadcast
- Step 2: Identify 172.16.0.0 as the lowest subnet ID and 126.96.36.199 as the highest subnet
broadcast address, defining the low and high end of the range that the summary
- Step 3: With all four masks as /24, subtract 1 from 24, so the initial value of /P to try is
Step 4 requires some math that gets repeated until the best summary is found. For the first time through the math at Step 4, use the /P calculated at Step 3 (in this case, /23).
Each time the calculation in step 4 does not result in finding the best summary route, repeat the math with a shorter prefix length mask, until the best summary is found. This initial pass through step 4 uses subnet ID 172.16.0.0 and mask /23. At this point, you do not even know whether 172.16.0.0 would be a subnet number when using mask /23, so do the math as if you were trying to calculate both the subnet number and broadcast address.
The calculation shows /23: subnet 172.16.0.0, broadcast 172.16.1.255
At step 4 part A, you need to compare the newly calculated subnet address range with the range of addresses in the original subnets, as identified in Step 2. The new potential best summary route doesn’t include the entire range of addresses for the original subnets. So, at step 4 part B, subtract 1 from the prefix length (23 – 1 = 22) and start step 4 again, with a /22 mask.
At the next pass through step 4, again starting with the lowest original subnet ID (172.16.0.0),
and using the current prefix /22, calculating the subnet ID and broadcast, you get:
/22: subnet 172.16.0.0, broadcast 172.16.3.255
Back to step 4 part A, this range exactly matches the range shown in the first figure, so you have found the subnet and mask to use in the best summary route: 172.16.0.0/22.