> Topology Overview:
- Routers RG 31/32 - peering routers
- Routers RR-10 / RR-20 - route reflectors
- Ps and PEs - core and customer connectivity
BGP CONCEPTS
> Pourpouse of BGP:
- interconnect between different AS (interdomain communication)
- widely used in Service Provider, Large Enterprise and Datacenter enviroments
- multihomed customers
- scalable
- created with stability in mind
> Main considerations
- trust one one - eBGP filter advertisment in and out
- policy based
- multiprotocol support IPv4/6, VPNv4/6 and VXLAN
- reliable updates
- triggered updates only
- uses rich metrics/attributes
> Internal BGP
- neighbourship between two nodes in the same AS
- AS-PAth is not updated when sending to iBGP
- BGP Split-horizon is used to prevent the loops, iBGP updates are not forwarded to any other iBGP peer
- BGP next-hop is not changed
- used to carry partial of full table of internet routing prefixes
> External BGP
- neigborship between two nodes with different AS numbers
- when router receives a route with its own AS in the AS Path the information is discarded (loop prevention)
- used to exchange prefixes with other ASes
- implements routing policies
> RFC 4456 / Route Reflectors
- removes the need for Full Mesh IBGP
- loop prevention is done using non-transitive attribute called CLUSTER_LIST and adding it is own ID to it
- when router receives update which CLUSTER_LIST contains router's own cluster ID the update is discarded.
- By default BGP Router ID value is used for cluster ID - has to be 32bit and can be changed i.e.: 0.0.0.1
- Multiple cluster IDs (MCID) feature allows to assign per-neighbor cluster IDs
> Possible scenarios of route reflection
- Between client and non-client
- Between clients in the same cluster (intra-cluster)
- Between clients in different clusters (inter-cluster)