Multiple routing processes with the same Admin Distance

[pkhatri]

Hi all,

If, for some reason, there are 2 or more routing processes on a router (not necessarily the same protocols) with the same Administrative Distance and the same prefix is learned by two or more of them, what are the rules for determining which route gets installed in the routing table ?

I do realise that this is poor design but the question is still valid...

Cheers,

Paresh Khatri.

[ankurbhasin]

Hi Paresh,

Before checking the admin distance it will check the longest prefix match and that route will be installed in routing table.

HTH

Ankur

[pkhatri]

Ankur,

Yes, I do understand that very well but if you read my question, you will see that I stated "same prefix", where I imply the same prefix and the same length....

Paresh.

[ankurbhasin]

Hi Paresh,

I m sorry I missed that part in your original post but yes if the routes have the same prefix length and admin distance then metric will play the role.

HTH

Ankur

[pkhatri]

Hi Ankur,

I think you are still missing the point here... I'm asking about what happens when you have the same AD for potentially different routing protocols.. comparing metric isn't really gonna work.

Paresh.

[ankurbhasin]

Hi Paresh,

I am not sure why you say metric will not come into scenario.

What I am sure is if to reach any destination I have eigrp also with admin dist 90 and also I configure ospf also as 90 definetely both the protocols will use diff algorithm to calculate the final metric and that will definetely vary till the time you tweak that values also.

And the best metric will decide which route should be choosen to reach that destination.

Metric is the only last option to differentiate the routes if admin distance prefix length is same.

Ankur

[pkhatri]

Hi Ankur,

I'm not sure that your example makes sense... Extending your example, if the OSPF route has a cost of say, 10000 and the EIGRP route has a minBw=64k and delay=20000 microseconds, how do you compare the metric between the two ? What makes one metric better than the other ? How can you compare non-homogenous quantities like these ?

Paresh.

[ankurbhasin]

Hi Paresh,

EIGRP will not compare its metic with just the values but it will use the value to take out the cost which it will take using this formula

10000000

( --------- + total delay ) * 256

min band

Now whatever the value comes it will be its metric and it can then compre the value with the value for ospf metric.

Ankur

[pkhatri]

Ankur,

That's comparing apples and oranges and I have absolutely no doubt that that is incorrect. While OSPF and EIGRP both have 32-bit metrics (well, the composite metric for EIGRP is 32-bit), that does not mean that it makes sense to compare them. What if you were comparing OSPF and RIP ? Would you compare a 32-bit quantity with a hop count ? I think not...

I might wait for someone from Cisco to clear this up.

Thanks for your responses anyway.

Paresh.

[Richard Burts]

You are correct that it is comparing apples and oranges. And YOU are the one who created apples and oranges when you specified two different protocols with the same administrative distance. It may not make logical sense to compare two dissimilar metrics, but the conditions that you specify will cause that to happen.

That is why it would be such a bad design.

HTH

Rick

[Harold Ritter]

Rick is right when he says that it is not good pratice to set two different routing protocols with the same admin distance as it can lead to serious confusion.

Here how IOS handles it though. If the same route is learnt from different routing protocols, let's say ospf and eigrp and that the admin distance has been set to the same value, IOS will use the default AD for each protocol to determine which route is preferred, in which case the eigrp learnt route would win.

Hope this helps,

[pkhatri]

Thanks a lot. That clears up the situation for me... Unfortunately, I have seen this being used in a customer network and the results were not nice. So I was just after some rules for deterministic calculation of the preferred route.

Cheers,

Paresh.