EIGRP bandwidth

kpjungle

Hi,

Im having a bit of trouble with bandwidth on WAN links using EIGRP.

In the self-study, on page 105, they have an example of a multipoint scenario, with a hub, and 4 spokes. 3 of the spokes have a CIR of 256, and the forth one with 56. In multipoint, one solution is to take the lowest CIR and multiply by the number of VC's. I then understand it, that you apply this number (in this case 224) to the hub router, because it will then average the bandwidth out (on multipoint), this means that each VC gets 56 bandwidth. What confuses me at the moment, is that the picture of the 3 high-capacity VC's, state: CIR 256, and BW 224.

According to rule 3 from cisco, it states: "The bandwidth allowed for EIGRP on each virtual circuit must be the same in each direction. "

As far as I can tell, shouldnt the 3 high-capacity spokes then have a bandwidth of 56 each, in order to have the same EIGRP bandwidth on each end?

So, to sum it up, is the drawing just a bit off, or am i missing the point completely?

Thanks!

gojericho0

kpjungle wrote:

As far as I can tell, shouldnt the 3 high-capacity spokes then have a bandwidth of 56 each, in order to have the same EIGRP bandwidth on each end?

Thanks!

Yes, this would be correct. You would configure 56 * 4 = 224 on the hub side and 56 on each of the spoke routers. Like you stated this will make the bandwidth the same on both directions. The problem with this is that the higher capacity links would be under utilized for EIGRP traffic.

I do not have the book, but the 224 could be an alternative solution of using p2p links instead of multipoint in order to properly utilize the bandwidth. I'm not sure why the BW would be configured for 224 on the high capacity if it was multipoint unless there was a mistake. Have you downloaded the errata off of cisco press to see if it was a boo boo?

kpjungle

gojericho0 wrote:

Have you downloaded the errata off of cisco press to see if it was a boo boo?

Yeah, if it is, they dont know it yet

Thanks though!

Netstudent

kpjungle wrote:

Hi,

Im having a bit of trouble with bandwidth on WAN links using EIGRP.

In the self-study, on page 105, they have an example of a multipoint scenario, with a hub, and 4 spokes. 3 of the spokes have a CIR of 256, and the forth one with 56. In multipoint, one solution is to take the lowest CIR and multiply by the number of VC's. I then understand it, that you apply this number (in this case 224) to the hub router, because it will then average the bandwidth out (on multipoint), this means that each VC gets 56 bandwidth. What confuses me at the moment, is that the picture of the 3 high-capacity VC's, state: CIR 256, and BW 224.

According to rule 3 from cisco, it states: "The bandwidth allowed for EIGRP on each virtual circuit must be the same in each direction. "

As far as I can tell, shouldnt the 3 high-capacity spokes then have a bandwidth of 56 each, in order to have the same EIGRP bandwidth on each end?

So, to sum it up, is the drawing just a bit off, or am i missing the point completely?

Thanks!

Perhaps they are explaining the BW percentage that EIGRP could use for EIGRP traffic that is configurable. By default it is 50% of the arbitrary BW on the interface. I don;t think you would want the default of 50% on one end and then configure 150% on the other end.

APA

You've missed the point.... It is to allow for efficient use of EIGRP when you have a mix of high and low bandwidth links mismatched in a multipoint setup.

224 is stated so that the high-bandwidth links still allow EIGRP to use up to 50% of 224 and the reason for multiplying 56 * 4 is to come up with a figure that will help prevent saturation of the lower bandwidth link.... Remember just because you have stated 224 does not mean EIGRP will always use 50% of this (112Kbps) your only telling EIGRP that should it need the full 50% of a link then it can only use 50% of 224....

Now the correct method for configuring the above scenario would be to

a) Create a multipoint interface with only the 256 links and specify the BW as 256 * 3 for the multipoint hub interface then 256 on each of the Remote Interfaces so the BW matches the CIR (remembering EIGRP will only use 50% of this)

b) Create 1 x Point-to-Point interface for the 56Kbps link with the BW set to 56 so it matches the CIR of the link (Remembering again EIGRP will only use 50% of this if needed)

This setup ensures the all the links have BW and CIR matching and no need to worry about saturating lower bandwidth links as you no longer have a multipoint scenario with mismatching CIR's

Hope this explanation clears this up for you

kpjungle

A.P.A wrote:

You've missed the point.... It is to allow for efficient use of EIGRP when you have a mix of high and low bandwidth links mismatched in a multipoint setup.

224 is stated so that the high-bandwidth links still allow EIGRP to use up to 50% of 224 and the reason for multiplying 56 * 4 is to come up with a figure that will help prevent saturation of the lower bandwidth link.... Remember just because you have stated 224 does not mean EIGRP will always use 50% of this (112Kbps) your only telling EIGRP that should it need the full 50% of a link then it can only use 50% of 224....

Now the correct method for configuring the above scenario would be to

a) Create a multipoint interface with only the 256 links and specify the BW as 256 * 3 for the multipoint hub interface then 256 on each of the Remote Interfaces so the BW matches the CIR (remembering EIGRP will only use 50% of this)

b) Create 1 x Point-to-Point interface for the 56Kbps link with the BW set to 56 so it matches the CIR of the link (Remembering again EIGRP will only use 50% of this if needed)

This setup ensures the all the links have BW and CIR matching and no need to worry about saturating lower bandwidth links as you no longer have a multipoint scenario with mismatching CIR's

Hope this explanation clears this up for you

Well, yes. I know its because of the low bandwidth link. The four spokes each get an even share, in this case 56kbps each. This is because the hub multipoint automatically evens out the 224. However, what I take issue with, is why (according to the schematic), to rely on the 56kbps on one end, and then configure 224 on the other end of the three high speed links. That defy the "cisco rule" of having the same bandwidth command configured on each end.

EIGRP will utilize max 50% 56kbps on each VC (to not overload the low bandwidth link). This will happen even on the high bandwidth link in one direction.

From the high bandwidth links and to the hub router, EIGRP will utilize (per default) 50% of the configured bandwidth (in this case 224kbps on each VC).

So when EIGRP sends something from the spokes, it can do so with higher bandwidth, but EIGRP from the hub has alot less bandwidth. 28kbps in one direction, 112kbps in the other.

To me it seems to defy that third "cisco rule", no?

APA

Indeed it does... hence why it is not the correct way to configure this scenario.....

In the study guide Cisco then goes onto explain the correct configuration as I have explained in my last post which was to create a multipoint interface which interconnects the 256Kbps links then create a PTP interface for the 56Kbps link this allows you to follow Cisco's golden rule of BW matching CIR in both directions.

kpjungle

A.P.A wrote:

Indeed it does... hence why it is not the correct way to configure this scenario.....

In the study guide Cisco then goes onto explain the correct configuration as I have explained in my last post which was to create a multipoint interface which interconnects the 256Kbps links then create a PTP interface for the 56Kbps link this allows you to follow Cisco's golden rule of BW matching CIR in both directions.

Okay, now i follow your post. Ok, thanks

kpjungle

Ack.. It would seem im not done with this subject just yet

Can someone explain this to me:

If I have a multipoint interface with two peers. My access line speed to the cloud is 256. What would my EIGRP bandwidth usage for each peer be?

My assumption is that since EIGRP divides it into even parts it would be calculated this way:

256/2 (peers) = 128kbps per peer. 50% (standard EIGRP usage of bandwidth) of this is 64kbps.

So according to my assumption, EIGRP will use up to 64kbps on each of these links, is that correct?
(yes, still dont understand why the above-mentioned drawing is wrong according to my mind )

Plazma

That sounds about right.. your formula is correct.