Border Gateway Protocol (BGP)
AS numbers
Each ISP or company is uniquely identified by an AS number. The 16-bit number was extended by RFC 4893 to 32-bit
| Purpose | AS numbers |
|---|---|
| Reserved | 0, 65535 |
| Public IANA-assigned | 1 - 64495 |
| Documentation/examples | 64496 - 64511 |
| Private | 64512 - 65534 |
eBGP and iBGP
There are two BGP flavors:
- eBGP: BGP running between BGP peers in different AS
- iBGP: BGP running between BGP peers in the same AS
iBGP
- peers can be connected indirectly
- peers do not advertise routes learned from other iBGP peers
- requires full-mesh (for large topologies –> (route reflectors, confederations)
eBGP
- peers require direct L3 connectivity by default (
neighbor ebgp multihop nenables BGP to be established over multiple hops)
iBGP
iBGP requires full-mesh connectivity between all iBGP peers. To avoid (N-1)/2 links Route Reflectors (RR) and confederations were introduced.
Route Reflectors (RR)
![[HPE7-A08.md#^frame=V7I2T9Ro]]
Confederations
eBGP next hop behavior
| Scenario | Next hop |
|---|---|
| Directly connected eBGP peers | eBGP session interface |
| Multihop eBGP | eBGP session interface |
| Route reflectors (RR) | Unchanged |
| iBGP next-hop-self | |
| iBGP next-hop-self or advertise outside route inside the AS |
BGP route selection
We Love Oranges As Oranges Mean Pure Refrehment
| Attribute | Description | Command | |
|---|---|---|---|
| Weight | Attribute that can be specified locally per route. | set weight 100 or neighbor x.x.x.x weight 100 | |
| Local preference | Used within iBGP to determine the best exit to reach a particular route. | set local-preference 200 | |
| (Locally originated) | |||
| AS path length | |||
| Origin | Routes injected to BGP are preferred (IGP < EGP < ?) | ||
| MED | Set by ISP routers | set metric 200 | |
| eBGP over iBGP | |||
| IGP metric to next hop | |||
| ORIGINATOR_ID | (If ECMP is disabled) | ||
| CLUSTER_LIST | (If ECMP is disabled) | ||
| Router ID | (If ECMP is disabled) | ||
| Neighbor’s IP address |
Commands
show bgp ipv4 unicast neighbor to verify neighbor information
- BGP state
- AS
- Up/down time
show bgp ipv4 unicast summaryto get a summary of the neighbor informationshow bgp ipv4 unicast pathsto see routes learned via BGP
show bgp to verify the BGP routing table
The BGP routing table is different from the IP routing table. Only valid and best paths from the BGP routing table are added to the IP routing table.
| Symbol | Description |
|---|---|
* | Route is valid and next hop is reachable |
e | Route is from foreign AS or protocol |
i | Native BGP route from local AS |
> | Route is best path to reach destination |