ROUTE Notes – EIGRP Topology Stuff

Study Questions

  • How do you keep EIGRP from killing your WAN?

You can use the ip bandwidth-percent eigrp AS X command to limit the amount of bandwidth that EIGRP uses to update neighbors.

  • How does EIGRP calculate how much bandwidth it can use for each frame relay PVC?

By default, EIGRP takes 50% of the (sub)interface’s configured bandwidth (with the bandwidth command) to use for updates on NBMA (non-broadcast mutliaccess) networks like frame relay.  This value is divided equally among all the PVC configured on that interface.

  • Why should you use delay instead of bandwidth to manipulate EIGRP?

There are other mechanisms, like QoS, that use bandwidth, so changing that value would affect those mechanisms.  Only EIGRP uses delay.

  • What’s the difference between the feasible distance (FD) and the reported (advertised) distance (RD)?

Feasible distance is the EIGRP metric value after the router has added it’s own information like bandwidth and delay to the formula.  The reported distance is what a router calculates before it has added it’s own values.  Essentially, the FD of one router is the RD of the next.

  • What is an offset list?

An offset list is a way to artificially increment the FD  and RD of a route or set of routes.

  • You can add the load k-value into the metric calculation in EIGRP, but it’s not generally a good idea.  Why?

The load is constantly changing as traffic changes on an interface.  This would cause a constant stream of updates as traffic flows change.

  • How often does a router send its full EIGRP topology table?

When new neighbors come up, the neighbors exchange their full tables, but, from that point forward, only updates are sent.

  • When we talk about bandwidth in EIGRP, what are we actually talking about?

The bandwidth is actually the bandwidth of the slowest link between a router and the destination network.  This is what’s used in the calculations.

  • Assuming we’re using k1 and k3, what is the formula for calculating the metric?

metric = 256 * ( 10^7 / bandwidth [in kbps] + cumulative delay )

  • You’ve decided to use k2 in your metric calculations, so you add that to a router.  What happens to all the neighbors?

The neighbors all drop and start generating a “K-value mismatch” error.

  • What is a successor?  Feasible successor?

A successor is the EIGRP route for a particular network entry with the lowest metric.  This is the route that EIGRP submits to the routing table for inclusion.  A feasible successor (FS) is another EIGRP whose RD is lower than the successor’s FD; feasible successors can be used as an alternate path to a network if the successor goes away somehow.

  • Where would you run into split horizon issues with EIGRP?

Split horizon says that you don’t advertise a route over the interface on which it was received.  If you have a multipoint WAN link of some kind, routes from one spoke won’t be passed to another spoke through the hub.

  • In what unit is the delay directive?

Tens of microseconds (10 * usec).  That means that delay 1000 is 10,000 usec, or 10 ms.

  • How does EIGRP do unequal cost path load balancing?

You can set the maximum-paths value under EIGRP to set the maximum number of equal paths that can be used.  You also set the variance command there with a multiplier integer.  The variance is multiplied by the successor’s FD, and any feasible successor whose metric is less than this new number is considered equal cost.

  • What is an EIGRP stub router?

A stub router only receives routes via EIGRP and does not send them to other EIGRP neighbors.  Since all the other routers know a router is a stub, they won’t send query messages to the stub router if they’re looking for a route.  This will cut down on time waiting in active state.

  • What is “stuck in active”?

If a successor for a network becomes unavailable and there are no FSes, a router will query each of its neighbors for a new routes to that network.  If that router does not have a route, it will then ask its own neighbors, etc.  In the meantime, the original router is still waiting for everyone to answer (that is, the route is in the active state) and will wait patiently until that happens.  This could take a long time and result in a several-second outage to the queried network.

  • By default, which types of routes are sent to EIGRP neighbors from a stub router?

Connected and summary

What Command Was That?

What command…

  • …disables split horizon on an interface?

R1(config-if)#no ip split-horizon eigrp 1

  • …sets the delay of an interface to 10ms?

R1(config-if)#delay 1000  (remember the tens of usec unit)

  • …restricts the amount of bandwidth that EIGRP uses to 20% of the total bandwidth on an interface?

R1(config-if)#ip bandwidth-percent eigrp 1 20

  • …apply an offset list to interface F0/0?

R1(config-router)#offset-list ACL in OFFSET F0/0

  • …show the metrics of all the EIGRP routes a router has recieved for a network?

show ip eigrp topology

  • …shows the successors and feasible successors for a network?

show ip eigrp topology

  • …shows if a neighbor is a stub?

show ip eigrp neighbor detail

  • …shows the maximum-paths and variance values?

show ip protocols

Aaron Conaway

I shake my head around sometimes and see what falls out. That's what lands on these pages.

More Posts

Follow Me:

7 comments for “ROUTE Notes – EIGRP Topology Stuff

  1. June 17, 2010 at 6:04 am

    Aaron, I truly envy you, you are so disciplined…. I am stuck at learning – again 🙁

    Seems like I have to start all over (ROUTE) for the third time now…anyhow I wish you good luck with your studies

  2. June 17, 2010 at 7:57 am

    SIA is my favorite EIGRP problem. Practical experience is that an SIA route can result in a more than several second outage. If you have a really large EIGRP topology, say hundreds or thousands of routers, and then you have some route instability – say a flapping head-end WAN circuit – you can end up with outages lasting for as long as it takes you to find the router with the bad EIGRP topology database. This will manifest itself not as a global outage, but rather as certain source networks not being able to get to certain destination networks. From there, you clear that bad route out with “clear ip eigrp topology” (I believe, doing this from memory).

    Summary is that really large EIGRP domains are a bad idea unless you’re disciplined about keeping your query scopes in check using route summarization and stub routers.

    All of which has little or nothing to do with your study notes…SIA scenarios just popped in my head as I was reading through them. 🙂

  3. Mohammed Ali
    June 17, 2010 at 3:20 pm

    Hi Aaron, Good luck for your studies,

    I would like to ask you about this exam, is there any major diffrences between BSCI and ROUTE bcz I’m currently studying for my BSCI, and I hope I can do it before the end of next July. But in the same time in case of bad luck (not to pass the BSCI) I could also at least have a good preparation for the ROUTE exam through my current preparation for the BSCI,

    Thanks in advance.

  4. June 17, 2010 at 3:44 pm

    Sebasian: Good luck to you as well. Keep at it, and it will be over before you know it! Keep the blog up, too. 🙂

    Ethan: SIA deserves more than the 2 or 3 paragraphs it gets in the book. It’s good to get more information on it, though, since, if history repeats itself, my instance of the test will wind up being 80% SIA if I don’t learn it. Heh.

    Mohammed: According to the blueprints (links below), BSCI has mutlicast and IS-IS, while ROUTE doesn’t. ROUTE, however, has more planning details that BSCI doesn’t; that is, there’s a lot more “determine this” and “verify that” on ROUTE. That’s the part I’m worried about. 🙁

    ROUTE blueprint:
    BSCI blueprint:

  5. June 20, 2010 at 5:42 am

    There are two SIA stories:

    The original SIA problem: once you lose a route, you start querying; the route goes ACTIVE. If you don’t collect all the responses within the timeout window, you declare an emergency and DROP THE NEIGHBORSHIP with the router that failed to produce the reply. Usually this happens between core routers; a single flappy remote site can thus bring the whole network down.

    This was happening a lot in original EIGRP and prompted Cisco to redesign SIA behavior. Now SIA does not cause adjacency reset, which also means that bogus information can survive for a long time, sometimes even “forever” (until you clear the neighborships manually). We’ve had scenarios where an EIGRP bug would introduce weird information in EIGRP topology tables and that information would not go away until you’d do “clear ip eigrp …”. The truly hard part of this behavior is that it’s totally random and unpredictable.

  6. June 22, 2010 at 12:27 am

    Here’s a slightly longer explanation for using delay to control your EIGRP routes:

  7. Steven
    December 24, 2012 at 12:48 pm

    I’m using the following formula in EIGRP to load balance between 3 unequal cost paths targeting a ratio of 25/18/7:

    10^7/((Metric/256)-(Cumulative Delay/10)=BW

    Be advise that I’m only manipulating the BANDWIDTH not delay with a FD of 412160 to the target network.

    I would like to send 25 packets thru Fa1/0 with a DLY of 6100, 18 Packets thru Fa2/0 same DLY 6100, and 7 Packets thru Serial0/0 DLY 25000.

    Could you PLEASE show me how to achieve a 25/18/7 Ratio? I’m not getting this Ratio following your steps.

Leave a Reply

Your email address will not be published. Required fields are marked *