BGP Local Preference...can I?

[darkscreamz]

Hi guys,

Thought this might be a helpful place to figure out whether what I want to do is achievable!

At the moment, traffic destined from STH-1(AS1111) for 192.168.0.0 or 10.0.0.0 default travels via MDL-1(AS1111)->NTH-1(AS1111)->AS2222 (the red highlighted links), this works fine, however I want to try and load balance the traffic via the MDL-2->NTH-2 link as well. From what I've read regarding BGP Local Preference + route-maps this seems possible but I'm having difficulty getting my head around it? I've included another diagram below of what I want to happen

So as you can see looking at the red highlighted links, I'm trying to make traffic flow to 192.168.0.0 and 10.0.0.0 over both of the following paths to AS2222 (I want to avoid using the NTH-2->AS2222 link):

STH-1(AS1111)->MDL-1(AS1111)->NTH-1(AS1111)->AS2222

STH-1(AS1111)->MDL-1(AS1111)->MDL-2(AS1111)->NTH-2->NTH-1(AS1111)->AS2222

As I said this sounds doable via bgp local pref, I'm just having a hard time figuring getting it to work in my test lab. For me to achieve this would I need to say manually setup a higher local pref route map for 10.0.0.0 on MDL-2 so that traffic is sent to there instead (192.168.0.0 would continue to default via MDL-1->NTH-1 in that case?) and it should then go via NTH-2? Or am I making this to hard with bgp local pref involved?

Would love some opinions

[Giuseppe Larosa]

Hello Darkscreamz,

local preference is meaningful within a single AS

if NTH2 sets a better local preference on route 192.168.0.0 using a route-map applied inbound to eBGP session with the router in AS 2222 you will achieve your objective.

current best path is chosen probably by lowest IGP metric to next-hop so the other prefix will keep using the current path

local preference is checked before looking at lowest IGP metric to BGP next-hop this is the key point here.

Hope to help

Giuseppe

[Giuseppe Larosa]

Hello Darkscreamz,

reading more carefully your objective after Mohamed's notes (thanks for noting this, rated as deserved)

>> So as you can see looking at the red highlighted links, I'm trying to make traffic flow to 192.168.0.0 and 10.0.0.0 over both of the following paths to AS2222 (I want to avoid using the NTH-2->AS2222 link):

this cannot be done unless you accept to use NTH-2-AS 2222 link, because NTH-2 cannot advertise a BGP prefix that is not its own best path

if we make NTH-2 to prefer NTH-1 advertisement via iBGP session, the NTH-2 will suppress its own advertisement and the rise of local preference that should be done here and not in another router is useless.

As Mohamed has noted you will need to deploy route reflector server to avoid full mesh issues in iBGP.

What can be achieved as I've noted in my first post is to use NTH-2 for sending packets with destination 192.168.0.0 using  NTH2's eBGP session with AS 2222.

Hope to help

Giuseppe

[Mohamed Sobair]

I think your design is totally incorrect, an your second path toward AS 2222 will always be unused. This is due to BGP split horizon rule. although if you use a Route Reflectoe here, it will not be used.

I will suggest you another appropriat design with your current Scenario if you wish,

Mohamed

I