CISCO guru's...check me on this (EIGRP)

Factual Questions
1

Just want a quick check on an EIGRP routing question and the CISCO site is a pain in the ass so thought I’d check here first.

Here’s the situation. There is a core backbone networks of routers attached via T1’s, so that router A is connected to routers B and C via separate T1’s (picture a triangle with each let a T1). I’m obviously simplifying here. This backbone is set up using a series of /30 bit subnet masks, so, on the first leg you’d have 10.10.10.0 /30 (with the routers host addresses bring 10.10.10.1 /30 and 10.10.10.2 /30), 10.10.10.4 /30 (.5 /30 and .6/30) on the second leg and 10.10.10.8 /30 (.9 /30 and .10 /30) on the third leg.

First question: In the EIGRP table, wouldn’t it make sense to do:

router eigrp 10
10.10.10.0 0.0.0.255

On all 3 core routers, instead of:

router eigrp 10
10.10.10.0 0.0.0.3
10.10.10.8 0.0.0.3

On router A and so on (to lazy to write out the others, but hopefully whoever is reading this is following along)? The backbone addressing (i.e. 10.10.10.x /30) is only used on the backbone, so there aren’t any other 10.10.10 networks elsewhere.
Next question. Attached to each network is a layer 3 switch which is also part of the EIGRP network. Each switch represents a different facility, and each one uses a convention like this:

Routing switch A off Router A: 10.11.B.X /24, where B is a building VLAN number (i.e. building 10 would be VLAN 10, so the address for the building would be 10.11.10.0 /24).

Routing Switch B off of Router B: 10.12.B.X /24

Routing Switch C off of Router C: 10.13.B.X /24

Ok, the question is, wouldn’t it make more sense to simplify the EIGRP tables on the routing switches to be:

Routing Switch A:

router eigrp 10
10.11.0.0 0.0.255.255

Instead of (what they currently have)

router eigrp 10
network 10.11.10.0 0.0.0.255
network 10.11.20.0 0.0.0.255
network 10.11.30.0 0.0.0.255
network 10.11.40.0 0.0.0.255
network 10.11.50.0 0.0.0.255
network 10.11.60.0 0.0.0.255
network 10.0.0.0
Seems simple to me, but the network engineer here insists that you can’t do what I’m proposing (of course, he also insists that their system needs all these weird static routes, even when I showed him that taking them out doesn’t effect the system traffic). Anyway, if someone who knows CISCO routing would check me on this I’d appreciate it…I’ve been doing more with HP in the last 4 years, so my CISCO is a bit rusty.

ETA: Screwed up the second octet on in the second example…should have been 11 instead of 12

-XT

2

Don’t really know that much about EIGRP and CISCO routing protocols, but I find as an introduction:

http://www.petri.co.il/how-to-configure-eigrp-on-your-cisco-router.htm
*How do you configure EIGRP in the Cisco IOS?
To configure EIGRP in the Cisco IOS, just follow these steps:

1.Set the bandwidth on your interfaces using the bandwidth command
Router (config-if)# bandwidth XX

(where XX signify the bandwidth of the WAN connection)

2.Start the EIGRP routing process and specify your AS number
Router (config)# router eigrp AS

Where the “AS” in the above-mentioned command stands for Autonomous System number. This number should be the same on all routers.

3.Once you are through this stage, next step is to instruct the router in order to advertise the networks that are directly linked to it. You can do it with the help of command
Router (config-router)# network X.X.X.X

You should enter this command for each network directly attached to the router. However, you can, in a single statement, enter the supernet of a group of subnets. When you do that, the router will automatically identify the subnets.

Optionally, you can configure the amount of WAN link bandwidth that an EIGRP router will use with this command:

Router(config-if)# ip bandwidth-percent eigrp XX*

the key seems to be Where X.X.X.X is the network id of a network that is linked directly to the router.

I suppose the problem is that for example, if A somehow gets a packet addressed to B on the port to C it will assume it should be able to route it through a locally atatched port, and will not be able to.

The second question seems logical - if 10.11/12/13.B.X networks are all direct-attached to each switch, the aggregate table should work. Whether it really does, only experimentation can verify. The question is - what does the connection between, say, Router A and Switch A use? If it is say, 10.11.0.X then the routing may get confused? Packets from Router A for the switch A (10.11.0.2, I assume?) would be routed to Switch A instead of being directly sent to it. Should still work, no harm done…

3

If I am understanding you correctly for your core router I would just set a static route to 10.11.0.0/16 to B, and another to 10.12.0.0/16 to C. At each site set a default route pointed to A. This removes routing updates over your T1 links. This assumes you have no links between B and C.

For a simple EIGRP config you should just use the following on all routers.

eigrp 10
network 10.0.0.0

The router will distribute the correct mask in the routing updates to it’s neighbors. This will work if you do not have any other networks that use 10.0.0.0/8 addresses that are not part of your EIGRP AS. From your list this is what you are doing already as the last network entry is 10.0.0.0. All the other entries are not needed unless you take out the network 10.0.0.0 line.

The only reason to specify all the subnets individually is if there is a network you do not want to advertise EIGRP. So if you had 10.10.0.0, 10.11.0.0, 10.12.0.0 in EIGRP and you do not want EIGRP to run on 10.13.0.0 then you would use a network command for each /16.

You will want to be carefull making changes to routing if you have alot of extra routing commands in your configuration (specifically the static routes you mention). The biggest help I found when cleaning up our routing was the command - “reload in” This will let you set the router to reload in x minutes if you do not cancel the reload command. So you can do reload in 15 then make your changes. If you cause a problem with routing that drops your connection the router will reboot, and go back to the saved configuration. If everything is working OK, cancel the reload, and save your config.

Finally if you do use static routes please name them. It is very frustrating to find a static route on a router that nobody knows anything about.
-Otanx

4

Hadn’t seen this thread updated and had kind of forgotten about it. I want to thank both Otanx and md2000 for their comments.

Exactly. In fact, that’s what I ended up doing (as well as getting rid of pages of static routes). My guess is that this network, like others I’ve seen, has sort of evolved over time. Originally, someone built a core network with certain design parameters, then later people (who, I suspect, didn’t understand how the original was working) added onto it.

I’ve actually not used this command before and I can see how helpful it would have been. What I ended up doing was white boarding out the core network and then expanding it to all the building devices (both layer 2 and 3), then mapped out the static routes. In just about every case there was no reason for the static routes that I could see…defaulting routing took care of most of them, and the rest were being propagated via EIGRP in any case.

Couldn’t agree more! My guess is that at some point they had a main trunk line go down, but there was some issue with convergence of the network, so someone added some static routes in to redirect the traffic…and then never removed them. This probably happened multiple times, and since it ‘solved’ the problem before, they just carried on (add to that there were probably multiple people involved at different times).

In any case, I’ve been able to massively simplify routing on this network, though some of the VRT Lite stuff on one of their spur networks is still giving me fits (it’s not working and I’m not sure why, as I’m not all that familiar with VRT Lite).

I appreciate the comments from both of you…thanks muchly!

-XT