Here is a printscrint of the output, sorry for the screenshot, It's from a PDF and can't copy it.

And yes :P I did it in a rack so it's not a missprint haha, in real devices it does the same thing.

What I'm looking for is some kind of rules as to when does it put a 0. For example, in the output the 67.7 router is a EBGP peer, but ive seen the same think (no MED, or at least not showed) from IBGP peers.

thanks in advance!

Hello Omar,

by looking at your sh ip bgp example we can make some notes:

advertisements with MED=0 are the result of the usage of network command on originating router.

We can notice the origin code i (rightmost character in the line).

instead the lines that refer to advertisements with a missing MED attribute have origin code ? so they are the result of redistribution into BGP of another routing protocol (or connected or static).

Here best paths are chosen based on the shortest AS path (> symbol).

Hope to help

Giuseppe

Thanks Giuseppe, that's what I though too, but I've seen that this is not consistent, look at the following output, where some eBGP routes have the MED=0 and some does not:

Hello Omar,

I wonder if a route-map setting origin to i has been used to create the eBGP routes with origin i and missing MED.

This is technically possible (set origin igp in route-map config).

But it looks like we cannot say anything definitive here.

I can say that MED can miss on a great quota of BGP routes in a real BGP table.

So, you cannot be sure that MED is present in a line of a sh ip bgp.

Hope to help

Giuseppe

I solved this trough trial&error, made some scenarios and here is the resulting rules:

1.- If the prefix is comming from a iBGP peer and the metric is not set, the "show ip bgp" shows MED = 0.

2.- If the prefix is comming from a eBGP peer, the metric is not set, and that eBGP peer is the originator of the prefix (by means of redistribution or network statement), the "show ip bgp" shows MED = 0

3-. If the metric has been modified, then for every prefix from (iBGP and eBGP) the MED is shown.

3.- Else, MED = null

Thanks for the help!

The problem with a MED value of zero is that it wins in most cases, which may not necessarily result in the best path (or preferred path). That is because if MED is inserted into BGP updates from external peers, the number is more often than not >0

I question why iBGP updates even contain a MED value at all. Its purpose it to hint to external neighbors about preferred inbound paths.

Ajaz

Hi Ajay,

I question why iBGP updates even contain a MED value at all. Its purpose it to hint to external neighbors about preferred inbound paths.

---->1 use for this is when you are using Multihoming with single ISP,since MED vlaue do not cross the AS others router in the
AS come to know about the exit point for the destination network ( customer ) through that MED which is used by IBGP ,you could also use AS prepend to acheive this but when you have multihoming with single ISP MED is a better option since the memory and processing required for AS prepend routes are more.

Chao

Vishwa