BGP Route Advertisement Rule | Route Reflector (RR) |Confederations
Summary Insights:
- In our previous posts, we have discussed BGP packets, states, and attributes.
- Today, we’ll move on to Route Reflectors (RRs) , their function, the roles of clients and non-clients, and the advertisement rules followed in route reflection.
- Finally, we’ll also cover different vendor configuration examples (Huawei, Cisco, etc.) and touch on BGP Confederation as another method for scaling iBGP.
BGP Route Advertisment Rule
After BGP establishment b/w BGP Peers then below rules are using for the router advertisement.
- The BGP router learns any routes from IBGP peer then it will advertise to the EBGP Peer only not IBGP.
- The BGP router learns any routes from EBGP peer then it will advertise to EBGP & IBGP Peer both.
NOTE:
Routes learn from IBGP peer will not advertise to IBGP peer to avoid loop inside Autonomous system.
IBGP Routes will not advertise to IBGP peer (split horizon rule)
Learn from EBGP peer will advertise to all IBGP peers
As per above rule we need to establish full mess connection for the IBGP peers in Autonomous system. If there are n routers in AS then n(n-1)/2 connection will be there. To avoid huge numbers of connection Router reflector & Confederations concept are used.
Router Reflector (RR):
As name showing that its reflect routes & to solve the problem of full mesh & split horizon rule in IBGP peering. There will be one main router called reflector & others are client & non-client.
All client & non-client will establish peering with Router reflector (RR) Router.
Non-client will be peer with RR & All other non-clients.
Client Router:
Client will establish & advertise routes to Route reflector (RR).Then RR advertise these routes to client & non-client. Client router will not establish BGP peering relationship with other Clients.
Non-Client Router:
Non-client router has normal BGP peer relationship with RR Router, means that Non-client routes will not advertise to non-client router. In other words there is non-reflecting peer relationship b/w RR & Non-cleint.
Note:
Router reflector will not change next-hop of received routes it’s just reflect the same routes to clients.
Advantage of RR:
- Reduce the BGP peering configurations.
- Avoid full mesh connectivity.
- Improve device performance due to above steps.
Disadvantage of RR:
- Only reflect best routes due to this multi path traffic carrying eliminated.
Types of Route Reflector (RR):
There are two main two types of Router refector.
IP RR: where IPv4 routes are reflecting its called IP RR.
VPN RR: In MPLS ISP Backbone network to reflect VPNV4 routes called VPN RR.
Above same concepts used for IPv6.
Router Reflector Rules:
Below are the Rule during reflection routes.
- Routes learn from client router will advertise to both client & non-client
- Routes learn from non-cleint will advertise to client only.
- Routes learn form EBPG peer will advertise to both client & non-cleint.
RR Configurations:
Route reflector configuration only is done on RR router under BGP not on client & non-client.
Configurations Template Of BGP Route Reflector:
Cisco IOS – Route Reflector
router bgp 100
bgp cluster-id 1
neighbor 10.1.1.2 remote-as 100
neighbor 10.1.1.2 route-reflector-client
Huawei (VRP) – Route Reflector
bgp 100
router-id 1.1.1.1
reflector cluster-id 1
peer 10.1.1.2 as-number 100
peer 10.1.1.2 reflect-client
Juniper – Route Reflector
set protocols bgp group IBGP type internal
set protocols bgp group IBGP local-address 1.1.1.1
set protocols bgp group IBGP cluster 1
set protocols bgp group IBGP neighbor 10.1.1.2 peer-as 100
set protocols bgp group IBGP neighbor 10.1.1.2 route-reflector-client
BGP Confederation
This is another method to resolve the split horizon or full mesh BGP connectivity issue.
🔹Confederation will divide an Autonomous System (AS) to Sub-AS,s
🔹Inside Confederation full mesh or RR conept can use.
🔹IBGP will use inside sub-AS
🔹EBGP peer will be there between Sub-AS,s
🔹Best to use in Big ISP Network
🔹Different AS number for each sub-AS & One main AS number for outside connectivity
