ip bandwidth-percent eigrp

ccnpninja

Hi,
I read that the ip bandwidth-percent eigrp [as] [percent] command should be issued on point-to-point interfaces to limit the amount of EIGRP traffic.
So the bandwidth command will only be used for route metric calculations?

astorrs

Yup, the higher level routing protocols just use bandwidth for route calculations, for EIGRP it's the formula:

metric = [K1*bandwidth + (K2*bandwidth)/(256 - load) + K3*delay] * [K5/(reliability + K4)]

The default values are K1, K3 = 1 and K2, K4, K5 = 0; and the last part (* [K5/(reliability + K4)]) is only used if K5 > 0.

scheistermeister

astorrs wrote:

Yup, the higher level routing protocols just use bandwidth for route calculations, for EIGRP it's the formula:

metric = [K1*bandwidth + (K2*bandwidth)/(256 - load) + K3*delay] * [K5/(reliability + K4)]

The default values are K1, K3 = 1 and K2, K4, K5 = 0; and the last part (* [K5/(reliability + K4)]) is only used if K5 > 0.

To add to that the most common way I have seen ti make one route preferred over another (with just EIGRP) is to simply change the delay since both bandwidth and delay are static configurable variables.

astorrs

scheistermeister wrote:

To add to that the most common way I have seen ti make one route preferred over another (with just EIGRP) is to simply change the delay since both bandwidth and delay are static configurable variables.

I used to do that all the time when I was doing networking in a partially meshed EIGRP environment.

There are 4 other ways mentioned here if anyone else is interested.

cisco_trooper

ccnpninja wrote:

Hi,
I read that the ip bandwidth-percent eigrp [as] [percent] command should be issued on point-to-point interfaces to limit the amount of EIGRP traffic.
So the bandwidth command will only be used for route metric calculations?

Correct. The bandwidth command is only used in calculating routing metrics. I have seen people try to use this command to actually limit the bandwidth of an interface, which is not what it does. It does NOTHING to the actual speed of the interface...

dtlokee

EIGRP will use only the lowest bandwidth along the path to the destination so often changing the bandwidth may not have an effect. Delay is a better choice.

ccnpninja

thanks guys,

metric = [K1*bandwidth + (K2*bandwidth)/(256 - load) + K3*delay] * [K5/(reliability + K4)]

do I need this formula for the BSCI exam?

networker050184

I'd suggest you know it. You should also know which values are used by default.

ccnpninja

I checked CBT Nuggets, Jeremy says we don't need the complete formula to be memorized.