how bgp selects is atrributes

[carl_townshend]

Hi all

With BGP, how does it select the best path, I know there is a long list of attributes, weight, local pref etc, But say I had 2 different links to a destination, a 10 meg link, and a DSL, would BGP always pick the higher bandwidth link, like eigrp or ospf ? is link speed one of the attributes? I gather weight, local pref, MED etc are all at standard settings unless you change them?

cheers

[andrew.prince]

Read the below URL - for the best technical explaination:-

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094431.shtml

[carl_townshend]

Hi there

I can see an attribute called Metric, what im getting to is that if we had the same prefrix advertised by 2 seperate routers with different bandwidths, assuming all attributes are default, would it pick the one with the highest bandwidth ? whats the attribute called for this, is it metric ?

[andrew.prince]

BGP does not perform path selection on circuit bandwidth. 

Metric or otherwise know as MED (multi-exit discriminator) is an optional nontransitive attribute. MED is a hint to external  neighbors about the preferred path into an autonomous system (AS) that  has multiple entry points. The MED is also known as the external metric  of a route. A lower MED value is preferred over a higher value.

[carl_townshend]

so basically, you have to manually tune it to use the optimal path rather than it select its best path normally ?

[andrew.prince]

I suppose so - but I would be surprised if you are receiving routes from both ISP's with the same AS Path count to a specific destination.  If you have a 10mbs circuit and a DSL circuit, I would suspect the provider would not advertise routes of the DSL circuit with a better AS path......but you never can tel!!

[Vasileios Bouloukos MSc, ITIL, CCIE]

I would strongly recommnend to manually tune the BGP attributes.

It will be much easier for you to define the traffic flows and you will not depend on the random BGP attributes of the received routes. Finally, you can also troubleshoot easier any routing issue since you will know the routing flows from your router to the internet. It is not difficult to modify the BGP attributes and you can find many examples to the web....and any time we are here to help you...

Hope that helps,

Vasilis

[Marwan ALshawi]

BGP sometimes referred to as a Path vector protocol which dose not consider bandwidth of the interface/path in the path selection

and as suggested above by other posters you need to use manual configurations and BGP attributes manipulation to get it working in the way you want

if you want to make your routing/BGP to be more intelligent and network aware you might consider Cisco PfR but this might add some complexity to the support if its not designed/configured correctly,  but it is an option too

hope this help

[nkowlahar]

BGP is a distance Vector protocol.  By default it will take the path with the least number of hops reguardless of bandwidth availabiltity on a pipe.  There are many ways to modify BGP announcements (weight, local pref, med, prepend.. etc.. ).  Givin your description above, in my opinion the quickest and easiest for mulit-homed end-users is to prepend the AS-path.   Basically making one route to that particual network look further away by introducing extra hops wihin the AS path.  

http://www.cisco.com/en/US/docs/ios/12_3t/12_3t11/feature/guide/gtbgpdas.html