Pages

3.27.2013

EIGRP Auto-Summarization

Wait... Auto-Summarization!?! Oh Yeah

Done, end of post!

Not quite. In the real world this is as far as you go, but for study purposes and the lab you might want to understand what EIGRP (or any other protocol) is doing when auto-summary is enabled. Or when you hire that new CCNA who thinks he knows everything and doesn't apply your EIGRP template because he is above templates. You then understand what effect he is placing on the network and can smack him in the face with a shoe! Fine... don't hit him with a shoe. Just calmly explain why auto-summary is bad.

So lets dig in.




Cisco's EIGRP Auto-Summarization automatically summarizes routes each time they pass a border between two different major networks. Cisco.com

What the hell does Cisco mean my "major networks"?

You have to think back to when subnet classes were all the rage. Your major classes or networks are:

Class A: 0.0.0.0 - 127.255.255.255 mask: 255.0.0.0
Class B: 128.0.0.0 - 191.255.255.255 mask: 255.255.0.0
Class C: 192.0.0.0 - 223.255.255.255 mask: 255.255.255.0

So when a EIGRP router learns multiple routes (within the same major network) from its neighbors, it will summarize these routes into a single summary and advertise the summary out to its neighbors. EIGRP can only advertise EIGRP routes installed in the routing table, so in order to advertise the summary the router must first create a route to the summary address which is then pointed to null0 in the local routing table.

Well whats wrong with that? Summarization is a good thing, it keeps your routing table clean. Well yes and no. Lets lab it up and see!!


We have a simple network here with all routers in AS 10. R1 is simulating multiple networks on its loopback interfaces and will advertise them to R2 and hopefully on to R3

Its a base config for all three routers:

Router 1:

Router 2 and 3:


We can see R2 is now neighbors with both R1 and R3.


R2 is also learning all loopback networks via R1 as a summary.


R3 looks similar to R2


So we should now be able to ping the loopback networks of R1, Right???


YAY we can. Everyone is happy and all works!!

So wait... why the hell did you just waste all my time for when everything works?

The issue falls into place when another network, which sits in the same major network or class as the loopback networks, is installed on another router. Notice how R1 has a EIGRP (D) route installed for the summary address to the loopbacks? Where is that pointing? Thats right Null0, just what we explained would happen earlier.


Now, When R2 adds the loopback network 1.1.22.0/24 everything falls apart.


Lets ping R1s loopback network from R3 again...

What the crap!?! Looking at R3s routing table nothing changed. R3 still points to R2 for all the 1.0.0.0/8 via 10.0.23.2. Hmmmmm...


Lets see whats going on over at R2


There it is! Just like R1, R2 installed an EIGRP summary route to Null0 for the whole 1.0.0.0/8. So when R3 sends the ICMP packets to R2, R2 just dumps them in the bit bucket via Null0. Not good.

You will notice all three routers also installed an EIGRP summary route to Null0 for the whole 10.0.0.0/8 subnet. EIGRP will do this for each major subnet or network it learns and advertises.

Obviously the resolution to this issue is to disable auto-Summarization. But I hope this helps you understand how EIGRP auto-Summarization handles routes.

Thanks!!