BGP Notes

Always use: update-source loopback0 for iBGP.

When a prefix is received from an eBGP neighbor, it must be advertised to both eBGP & iBGP neighbors.

When a prefix is received from an iBGP neighbor, it is advertised ONLY to eBGP neighbors and NOT to any iBGP neighbors. This requires that all iBGP routers must be fully meshed or the use of route-reflectors or confederations.

The synchronization rule only applies to iBGP learned prefixes. It states that an iBGP learned prefix cannot be considered best unless you have a matching IGP route for that BGP prefix.

There are only 3 ways to originate prefixes in BGP:

  1. Network statement
  2. Aggregate statement
  3. Redistribute an IGP into BGP

When iBGP speakers are NOT fully meshed and there is no route-reflector or confederation, any prefix learned from an iBGP neighbor will only be advertised to eBGP neighbors and not to any iBGP neighbors.

BGP only advertises what it considers the best path. If the iBGP path is not synchronized, it's not included in the best path calculation.

Implications:

  1. NO SYNC or
  2. Synchronize by redistributing BGP into the IGP at the router that introduced the BGP route.
  3. Redistribute eBGP into IGP at the border router.

Forwarding is always done to the NEXT_HOP of the actual route and not from the router that originated the prefix.

To advertise a route:

  1. BGP must know the route through network command/redistribution
  2. Network advertised must be in routing table [unless no sync]
  3. There is a maximum of 200 network statements.
  4. If more than 200 routes, the routes must be redistributed.

Always set the bgp router-id ie: bgp router-id 223.2.2.2

The ospf router-id must be the same as the bgp router-id for redistributing the routes from ospf to bgp. It needs to be done manually if a route-reflector is involved or it won't work.

The BGP Router-id is used with the synchronization and in iBGP the BGP Router-id is used as tie breaker for the BGP path selection.

  • Just because the next hop is pingable, it does not mean it's a valid next hop.
  • Next hop addresses that are reachable only via a default route aren't valid.
  • Next hop addresses that are reachable only via another BGP route are also not valid.

By default, BGP does not accept subnets redistributed from Interior Gateway Protocol (IGP).

To advertise and carry subnet routes in BGP, use an explicit network command or the no auto-summary command.

The no auto-summary command is NOT needed if there is no redistribution from IGP into BGP.