administrative distance values assigned to protocols?

[ocicat002]

In all that I have read in preparation for the CCNA, I have not seen any explanation about *why* the various protocols are assigned the values stated. RIP has a value of 120, OSPF has a value of 110, & EIGRP has a value of 90. Numerically, this ranks OSPF similarly to RIP. However, EIGRP is significantly more "trustworthy" than either of these protocols, but *how* are these values determined?

Any insight anyone can share would be greatly appreciated. Thanks.

[Richard Burts]

James

The concept of Administrative Distance is a Cisco concept. Cisco assigned the values to reflect a system to weight the various protocols. As you state in effect the AD is a measure of trustworthiness. And in the view of Cisco EIGRP is more trustworthy than OSPF which is more trustworth than RIP.

What these valuse assert is that the view of the metric developed by EIGRP is more sophisticated (since it includes several factors in its calculation) and informative than OSPF (which measures a single factor) and that OSPF is more sophisticated and informative than RIP. If you share this view then use the default values. If your view is different then you can set different values for AD.

HTH

Rick

[ocicat002]

> What these valuse assert is that the view

> of the metric developed by EIGRP is more

> sophisticated (since it includes several

> factors in its calculation) and informative

> than OSPF (which measures a single factor)

> and that OSPF is more sophisticated and

> informative than RIP.

Understood. I suspect that these AD values were determined heuristically, but how? It has to be more than simply pulling numbers out of the air.

> The concept of Administrative Distance is a

> Cisco concept.

Yet isn't the purpose of AD values to allow comparison on a normalized scale? Otherwise, there isn't a quantifiable way to compare RIP hop counts to EIGRP bandwidth numbers. I'm looking for the rationalization for the values used on this AD scale.

[Richard Burts]

James

I do not believe that it is a normalized scale. It does not matter how much different 2 values are, it only matters which one is more attractive (lower).

And there is no meaningful comparison of hop count to EIGRP bamdwidth. It would be like trying to compare 2 objects knowing only that 1 object was 15 inches long and that the other object weighed 25 ounces. how do you normalize that?

side note: if you assume that all interfaces are the same bandwidth (which also implies that all interfaces have the same delay) then calculation of the EIGRP metric is exactly comparable to counting hops in RIP and produces exactly comparable metrics and would produce exactly the same choice of best path. But as soon as you introduce any variation in bandwidth then the metric of EIGRP is able to distinguish differences which are invisible to RIP and EIGRP would produce path choices which we would believe are better paths than what RIP would choose. Just for an example (admitedly slanted) if A is trying to get to B and has 2 choices: 1 choice is 3 hops each of which is 100 Mb and the other choise is 2 hops each of which is 64Kb, which would you want to use? (RIP chooses the least number of hops while EIGRP chooses the highest bandwidth and lowest delay)

HTH

Rick