Route summarization means replacing a series of routes with a summary route and mask which covers all of them. Most of the times, it’s not the case for small networks as the inter connecting devices have small routing tables, with just a few routes that don’t make problems.
When the network tends to be big and you have multiple inter connecting devices with tons of LANs and network segments split all over the place, then the problems start to arise. In this case, routing tables become bigger too, being composed by many lines that consume more memory and more time for router or L3 switch to be searching the routing table when needed to route a packet.
This is where route summarization comes and saves the day. Route summarization is a technique that helps network administrators in reducing the size of the routing tables.
For routing protocols like RIP, IGRP, and EIGRP, this technique also helps in reducing the update packet size because where it’s possible, instead of three or four routing entries it would have to send just one for example.
When the network administrator decides he will use route summarization he should first inspect the routing protocols working on the network. For route summarization to be possible, the routing protocols should support VLSM (variable-length subnet masks). RIP, EIGRP and OSPF, all of them support VLSM.
VLSM means in short terms, the use of multiple subnet masks in the same network. VLSM also reduces the waste of IP addresses on the network because if SLSM (static-length subnet masks) was used, it would be only one (the same) subnet mask on the network and many of the subnets would have IP addresses unused. VLSM prevents this by creating smaller variable networks based on how many host IP addresses are needed.
PPP is a WAN connection type also known as Point to Point Protocol. This protocol also stands for an encapsulation type available on WAN link for point to point connections between devices using TCP/IP stack.
On a Point to Point link we only need 2 IP addresses (one for each link) to configure the WAN link and we will show the use of route summarization with the help of some examples.
The use of a class C network with a subnet of 255.255.255.0 or /24 would be a total waste for such a connection when you only need 2 IP addresses out of the 254 usable. A subnet mask of 255.255.255.252 or /30 is more fit for this as you can see below on the WAN link with ip addresses for the interfaces of 126.96.36.199 and 188.8.131.52, both of them being a /30 subnet.
This is just a common example to display the actual use of the only 2 IP addresses we needed for the WAN link. I didn’t show the rest of the details behind a real network but you can imagine there are many more devices and equipments than 1 switch and 3 computers for a LAN.
Route summarization can be of better use if the network is designed with this technique in mind and you have the liberty to actual design it. If not, it’s still applicable but it will cost you more time and most probably you will have to do it on your free time, as you can’t remove the actual routes with no summarization to apply the technique after, without users noticing it.
As you can see I tried to keep a convention so we have R2 and the WAN links (usually this is the HQ in live environments) that start with 10.60, R1 with 10.50 and R2 with 10.55, both of them usually being Branch Offices or Remote Locations in live environments.
All of them except for the WAN links are /24 which actually means 255.255.255.0 and the WAN links have a subnet mask of /30 which means 255.255.255.252.
Now, how do we apply summarization ?
Let’s try to find a match for R3 networks. As we can see there, the IP addresses are in decimal format so we need to convert all three IP addresses to binary before doing anything else.
As you may already know, each IP address is formed of 4 octets, each octet is 1 byte which converted to binary would make 8 values of 1s and 0s, as you can see above. We have in here the last octet that for all the IP addresses is the same and we will ignore. The same applies for the first two octets too. The one we have to work with for the summarization is the one that differs, in our case being the third octet, so we have to find a pattern and common grounds for our summarization.
I used the red color so you can see what is the actual difference between the IP addresses while in binary. We have 6 bits which are the same, those marked with strike-out and the two left red ones that differ.
Based on this, we can find out the subnet mask because we know that all the first 22 bits must be the same to cover and match the same network, as you can see above in binary.
Note: Somewhere around here you can stop and just start adding the route to the routers but I would like to explain a bit more about the rest of the concept, if there is a chance you will understand this better… So here we go.
The subnet mask always counts for “1”s for network part and “0”s for the hosts part. so we can mark all the first 22 bits with a “1” and the rest of the bits to 32 with “0”s and you should have 11111111.11111111.11111100.00000000 (which is 255.255.252.0 or a /22 subnet mask) that covers for all the three networks and turns them into one super network, in other words a bigger one.
If we know the subnet mask, we can find the range as we know the IP address that we started with: network IP address 10.55.4.0, the subnet mask which is 255.255.252.0 and the ending IP address should be the broadcast which we can find if we set “1”s to all the host bits part. This means the broadcast IP address for this network can be found if we take the binary value of the network IP which is 00001010.00110111
.00000100.00000000 we count 22 bits and we let them as they are, then we change the left values to “1”s (32 bits -22 bits = 10 bits, 32 bits is the dimension of an IP address while 22 bits are the network part so we will get 10 bits as the host part). This would give us a broadcast address for this network of: 00001010.00110111 .00000111.11111111 (10.55.7.255).
How did this help us ?
Well now you can go and add to the routing table on R1 or R2 a network summarization of 10.55.4.0 with a /22 that covers all three networks of 10.55.4.0, 5.0 and 6.0 until a last address of 10.55.7.255 which is the broadcast address.
Instead of 3 routing table entries, you can have just one.
The rest of the example is made in the same way so I won’t discuss it and i’ll let you try it by yourselves. Do ask me if you don’t manage to do it.
RIP v2 and EIGRP route summarization can be done manually by being configured on the interface that will be doing the summary advertising. This is done by using “ip summary-address” command.
Also, the same routing protocols use auto-summarization on routes that will be advertised across classful network boundaries. This can be disabled by using “no auto-summary” command.
OSPF is more complex in this part, in such terms that to summarize routes from one area to another you have to use the “area range” command to summarize routes learned via redistribution as for the ASBR use the “summary-address” command.