- When generate summarized route, if AS_PATH in all subnets routes are the same, the summarized route will keep the same AS_PATH,
if subnet routes have different AS_PATH, in generated summarized route AS_PATH will be set as noll.
In order to keep AS_PATH track in summarized route, AS_SET option can be used in the command:
aggregate-address 188.8.131.52 255.0.0.0 summary-only as-set
- BGP will select the best route to advertise to the peer. it follows the best route selection policy here:
Local preference > AS_PATH > lowest Orignal code >the lowest multi-exit discriminator > eBGP route over iBGP > IGP metric to the BGP next hop > lowest router ID.
First of all, routes need to be the valid route before it is qualified to best route selection. A valid route means that Router has route path towards the ip of NEXT_HOP
- eBGP will update NEXT_HOP will advertise the route to the peer, but iBGP will keep NEXT_HOP unchange. NEXT_HOP can be modified with command
neighbor 184.108.40.206 next-hop-self
- Back door, if network 220.127.116.11 is learned from both eBGP and OSPF, since eBGP has lower AD, route learned from eBGP will be selected. This is not alway prefered. In that case,we can set backdoor as below
network 18.104.22.168 backdoor
Network 22.214.171.124 is treated as a local entry, but is not advertised as a normal network entry. After this, 126.96.36.199 will use the route learned from OSPF instead of eBGP.
- synchronization is enabled by default in order to avoid blackhole in the network.
it pretends a learned routes from being advertised to other peers if the same route cannot be learned from IGP route.
There are 2 ways to solve blackhole problem:
1, synchronization and redistribute all eBGP learned route into IGP route. While synchronizaton pretends a learned routes from being advertised to other peers if the same route cannot be learned from IGP route.
2, configure iBGP on each router of the network and all iBGP routers build a full mesh peer network.
This will bring high performance load if the network is too big and each router need run iBGP to maintain a big routing table. This problem is addressed by two ways: Confederation and reflectors.
- Default route
The default-information originate, redistribution from a different source, and network 0.0.0.0 are all similar in the resulting effect: they will inject the default route into BGP RIB and it will be advertised to all BGP neighbors. The difference is in the origin of the default route that is injected into BGP:
- default-information originate causes the default route to be artificially generated and injected into the BGP RIB, regardlessly of whether it is present in the routing table.
- Redistribution and network 0.0.0.0 will inject the default route into BGP only if the default route is currently present in the routing table, and additionally in the case of redistribution, if learned by a specific source protocol we are redistributing from.
The neighbor X.X.X.X default-originate is different in that the default route will be advertised only to this specific BGP neighbor and not to all existing BGP neighbors as with the previous approaches. The default route will not be present in the BGP RIB of the router that is configured with the neighbor X.X.X.X default-originate command and so it won’t be generally advertised to all BGP neighbors. At the same time, this command is similar to the default-information originate in that the default route is artificially generated and does not need to be present in the routing table.