cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1012
Views
20
Helpful
7
Replies

iBGP design problem

Hamed Karimi
Level 1
Level 1

Hey Guys

R1 and R3 are edge routers and have iBGP together. They have eBGP with R5. Link 1 is primary and if it goes down traffic which enters R3 should route to R1 and then to R5.

Here is the problem. If I dont include R2 in iBGP, then it will not know anything about R5. If I include it in iBGP, when link 1 goes down R2 will not advertise the routes received from R1 to R3 as all of them are iBGP. Any solution?

Capture.PNG

2 Accepted Solutions

Accepted Solutions

cofee
Level 5
Level 5

Hi,

You can configure R2 to be an iBGP route reflector and configure R1 and R2 as RF clients. If you don't want to run bgp on R2 and configure it as an iBGP route reflector then you can redistribute bgp routes into your igp, that way r2 would know how to forward traffic.

View solution in original post

Hi

It depends of what you want to achieve, if R1 and R3 are receiving the same networks from the external router (R5) you could configure BGP attributes on R1 and R3 in order to R2 can select the best path and have a backup. 

 

About

R1 and R3 are edge routers and have iBGP together. They have eBGP with R5. Link 1 is primary and if it goes down traffic which enters R3 should route to R1 and then to R5.

 

The traffic will be discarded by R2 (BGP Split horizon rule) if R1 want to receive that information but I dont see the reason to move traffic from R3 to R1 and then to R5, because R3 is an edge router (there is no LAN behind it), the idea here (I assume) if your LAN is connected to R2, R2 should select the best path and if it is down the traffic should be routed to other edge router as backup. 

 

 

For example: 

Imagine this config

 

Router 1 (Primary path)

 

router bgp 123

no syn

no au

bgp default local-preference 10000

neigh 10.15.0.5 remote 5

neigh 10.12.0.2 remote 123   (peer with R2)

neigh 10.12.0.2 next-hop-self

 

 

Router 3 (Backup path)

 

router bgp 123

no syn

no au

bgp default local-preference 2000

neigh 10.35.0.5 remote 5

neigh 10.23.0.2 remote 123   (peer with R2)

neigh 10.23.0.2 next-hop-self

 

 

 

 

R2 should prefer the path through R1 instead of R3 because it has higher local preference. It will change if LANs are behind R1 and R3 because the BGP split horizon rule will discard the packets if you dont have a full mesh o route reflector router. 

 

Also take in consideration my config above is just an example but there are proper ways to configure the BGP attributes filtering specific prefixes using PBR. 

 

Hope it is useful

:-)




>> Marcar como útil o contestado, si la respuesta resolvió la duda, esto ayuda a futuras consultas de otros miembros de la comunidad. <<

View solution in original post

7 Replies 7

cofee
Level 5
Level 5

Hi,

You can configure R2 to be an iBGP route reflector and configure R1 and R2 as RF clients. If you don't want to run bgp on R2 and configure it as an iBGP route reflector then you can redistribute bgp routes into your igp, that way r2 would know how to forward traffic.

e.ciollaro
Level 4
Level 4

There are many solution, for example you can

  • configure a full mesh topology (R2 has iBGP neighbourship with R1 and R3)
  • redistribute BGP in an IGP (not suggested)
  • configure a static default route on R2 having R1 as next hop
  • configure R2 has a route reflector 

The right choose depends on your architecture, your goals and your restrictions;  is this a real problem or are you just studying ? In the first case could you give us more details about the architecture ?

 

Bye

enrico

 

PS please rate if useful

 Static route on R2 will fix the issue but would create problem when primary link is back up. For static routes to work you would need to configure 2 static default routes pointing to R1 and R3, one  that points to secondary link needs to have a higher AD and track the primary link with an ip sla so once the primary link is down, primary static routes would be removed and back up static route with a higher AD will be installed.

 

 

Could be but the initial question was "traffic which enters R3 should route to R1 and then to R5." If traffic enter R3, R2 does not need a default to R3 because it is used just in case traffic as to go to R1 (I supposed that routes for returning traffic is present in R2 routing table). Of course a lot of details are missing here so may be my assumption is wrong and you are right

Bye,
enrico

What was asked and the topology that was shared your solution of using a static route on R2 and point it towards the secondary link will work and is correct. My comment was more towards a real world scenario that would involve LAN south of edge routers.

Hi

It depends of what you want to achieve, if R1 and R3 are receiving the same networks from the external router (R5) you could configure BGP attributes on R1 and R3 in order to R2 can select the best path and have a backup. 

 

About

R1 and R3 are edge routers and have iBGP together. They have eBGP with R5. Link 1 is primary and if it goes down traffic which enters R3 should route to R1 and then to R5.

 

The traffic will be discarded by R2 (BGP Split horizon rule) if R1 want to receive that information but I dont see the reason to move traffic from R3 to R1 and then to R5, because R3 is an edge router (there is no LAN behind it), the idea here (I assume) if your LAN is connected to R2, R2 should select the best path and if it is down the traffic should be routed to other edge router as backup. 

 

 

For example: 

Imagine this config

 

Router 1 (Primary path)

 

router bgp 123

no syn

no au

bgp default local-preference 10000

neigh 10.15.0.5 remote 5

neigh 10.12.0.2 remote 123   (peer with R2)

neigh 10.12.0.2 next-hop-self

 

 

Router 3 (Backup path)

 

router bgp 123

no syn

no au

bgp default local-preference 2000

neigh 10.35.0.5 remote 5

neigh 10.23.0.2 remote 123   (peer with R2)

neigh 10.23.0.2 next-hop-self

 

 

 

 

R2 should prefer the path through R1 instead of R3 because it has higher local preference. It will change if LANs are behind R1 and R3 because the BGP split horizon rule will discard the packets if you dont have a full mesh o route reflector router. 

 

Also take in consideration my config above is just an example but there are proper ways to configure the BGP attributes filtering specific prefixes using PBR. 

 

Hope it is useful

:-)




>> Marcar como útil o contestado, si la respuesta resolvió la duda, esto ayuda a futuras consultas de otros miembros de la comunidad. <<

Dennis Mink
VIP Alumni
VIP Alumni

your topology is not something you come across in real life  a lot.  I would let R1 and R3 form iBGP on the edge and use OSPF behind it. and use eBGP to R5

Please remember to rate useful posts, by clicking on the stars below.

Review Cisco Networking products for a $25 gift card