BGP

General Notes

AD 200 for iBGP, 20 for eBGP
Forms connection over TCP port 179 - does not need direct connection
Only advertises best routes, all others not advertised
Best-path Algorithm
1. Weight (Highest)
2. Local Preference (Highest)
3. Originated locally
4. AS_PATH (Shortest)
5. Origin code (IGP > EGP > ?)
6. MED (Lowest)
7. Path (eBGP > iBGP)
8. IGP metric (Lowest)
9. Route age (Oldest, only compared when routes from different eBGP peers)
10. RID (Lowest)
- We Love Oranges AS Oranges Mean Pure Refreshment
iBGP
- TTL of 255 by default
- Does not change next-hop by default - can be changed with neighbor <ip-address | peer-group-name> next-hop-self
- Uses split horizon¹ for loop prevention
eBGP
- TTL of 1 by default - can be increased with neighbor <ip> ebgp-multihop 255
- Can also use neighbor <ip> ttl-security hops <hop-count>
- Difference between above is that eBGP-multihop sets the maximum range allowed, whereas TTL-security sets the exact range
- If peering with different update source but not through additional hops, can use neighbor <ip> disable-connected-check
- Does change next-hop by default
Peer groups
- Can be used when multiple neighbors have the same requirements
- Use neighbor <name> peer-group to set up peer-group, then perform config under neighbor <name> <command>
- Peer groups applied to neighbor with neighbor <ip> peer-group <name>
- Can check with show ip bgp peer-group
- Peer groups are natively set up on the back end to limit CPU usage - only purpose is simplification of config
ECMP
- Can enable load-balancing with maximum paths <path#> command
- Only takes effect when many parameters on routes match - Weight, local preference, AS-PATH (both numbers and lengths), origin code, MED, IGP metric
Exact match of prefix in route table required for advertisement - configuring null route to prefix common
Public range for 2-byte AS: 1 - 64511, Private range: 64512 - 65535
- Also 4-byte ASs available now due to exhaustion of 2-byte ASs
Good idea to change update source to a loopback interface with neighbor <ip> update-source <loopback>
clear ip bgp * soft useful to refresh route table without tearing down neighborships
Selectively advertise default route with neighbor <IP> default-originate [route-map <CONDITION>]
Advertise a different AS to a peer than router bgp AS with neighbor <IP> local-as <OldAS> - useful in migration
neighbor <IP> fall-over can be used to immediately tear down session if IGP route to neighbor disappears

Features

Summarization

Primarily done with the aggregate address <prefix> <mask> command
- Only works for prefixes already in BGP table
- Router performing summary installs null route into table for aggregate
- Should be used with as-set suffix - this will include AS-PATH information and prevent loops/suboptimal routes
  - Inherits community values of any specific prefixes
  - For example, if one of four specific prefixes has a no-export community, summary with as-set will as well
  - These can be modified or stripped with attribute-map suffix
  - Can also use advertise-map suffix to select specific prefix which will determine attributes of summary
- By default does not suppress specific prefixes - summary-only at the end of command will
- Prefixes summarized in this way have ATOMIC_AGGREGATE attribute assigned
- Also assigned AGGREGATOR attribute specifying AS number and RID of aggregator
- Can selectively suppress specific prefixes with suppress-map - all prefixes in map will be suppressed:
```
ip prefix-list SUPPRESS_PREFIX 150.1.1.0/24
!
route-map SUPPRESS_MAP permit 10
 match ip address prefix-list SUPPRESS_PREFIX
!
router bgp 200  
 aggregate-address 150.1.0.0 mask 255.255.0.0 suppress-map SUPPRESS_MAP as-set
```
- Can also use unsuppress-map combined with summary-only to advertise specific prefixes
- Since this is per-neighbor it can be used for inbound path manipulation - you would advertise the same summary prefix to multiple neighbors and use an unsuppress map towards the preferred peer for inbound traffic:
```
ip prefix-list NET_1 permit 10.0.1.0/24 
!
route-map UNSUPPRESS_MAP permit 10
 match ip address prefix-list NET_1
!       
router bgp 200
 aggregate-address 10.0.0.0 255.255.252.0 summary-only as-set
 neighbor 155.1.37.7 unsuppress-map UNSUPPRESS_MAP
```
- Alternatively, can advertise a summary prefix in IGP with the network command - commonly done with a static null route

Route Reflection

Used to bypass full mesh requirement in iBGP
Configured with neighbor <ip> route-reflector-client on route reflector
Reflector must have full mesh if full reachability is required (unless using multiple reflectors)
Abide by following 3 rules:

Routes learned from eBGP peers can be sent to other eBGP peers, clients, and non-clients.
Routes learned from client peers can be sent to eBGP peers, other client peers, and non-clients.
Routes learned from non-client peers can be sent to eBGP peers, and client peers, but not other non-clients.

Functions like DR/BDR in OSPF by default - does not insert itself into transit path
next-hop-self command only affects eBGP routes when used with route reflectors

Confederation

Used to reduce full-mesh BGP requirements by splitting one public AS into smaller sub-as
Sub-as use eBGP advertisement rules between each other, removing full mesh requirement
Notable exception to eBGP behavior is that next-hop is not modified between confederation peers
Configured by using sub-as with router bgp command and using bgp confederation identifier <as> under router bgp
Other sub-as defined with bgp confederation peers <sub-as> under router bgp
Can be used in combination with route reflection

Communities

Activated for transport to peers with neighbor <ip> send-community
Made more readable with global config command ip bgp-community new-format
Applied to prefixes with set community <value1> <value2> ... <valueN> in route-map
To add communities without affecting existing, use set community additive <value1> <value2> ... <valueN>
Matched with community-lists - like ACLs, both standard and expanded (extended) versions
- Standard version uses 1-99, only permits or denies communities
```
ip community-list 1 permit 100:10 100:20
ip community-list 1 deny no-export
```
- Standard version uses AND logic with multiple communities in one line, OR logic with multiple lines
- Expanded version functions the same but adds regular expression functionality
Sample config of setting community:

ip as-path access-list 1 permit 60$
!
route-map SET_COMMUNITY permit 10
 match as-path 1
 set community 100:200
!
route-map SET_COMMUNITY permit 100
!
router bgp 200
 neighbor 155.1.45.4 send-community
 neighbor 155.1.45.4 route-map SET_COMMUNITY out

Sample config of matching community:

ip community-list standard 100:200 permit 100:200
!
route-map SET_LOCAL_PREFERENCE permit 10
 match community 100:200
 set local-preference 200
!
route-map SET_LOCAL_PREFERENCE permit 100
!
router bgp 100
 neighbor 155.1.13.3 route-map SET_LOCAL_PREFERENCE in

Well-known communities:
- no-advertise
  - Set with set community no-advertise
  - Signals to not advertise the prefix with this community to any peer
- no-export
  - Set with set community no-export
  - Signals to not export this prefix from the AS - can be advertised within AS only
- local-as
  - Set with set community local-as
  - Functions the same as no-export but also disallows advertisement between confederation sub-as
Deleting communities done with a community-list referencing community to be deleted, then using set comm-list <comlist> delete under route-map

Next-hop modification

Other than next-hop-self, you can also manually define next-hop with a route-map:

route-map SET_NEXT_HOP_FROM_R8 permit 10
set ip next-hop 155.1.58.8
!
router bgp 100
neighbor 155.1.58.8 route-map SET_NEXT_HOP_FROM_R8 in

BGP regexp

Can be used to search for specifc AS-PATHs in table - for example: show ip bgp regexp ^$
Above will show prefixes originated locally with the AS. Cheat sheet here.
Can be used to filter with ACLs with ip as-path: ip as-path access-list 1 permit _54$
Then use match as-path under a route-map to apply modifications

BGP backdoor

Created to prefer links via IGP over eBGP (raises AD of specific prefix to 200)
Configured with network <subnet> mask <netmask> backdoor under router BGP

Conditional Advertising

Allows advertising based on existence of another prefix in table
Configured with neighbor <IP> advertise-map MAP1 [non-exist|exist-map] MAP2
First map is prefix to advertise, second is prefix to monitor
BGP checks on existence of MAP2 every 60 seconds

Conditional Route Injection

Functions similar to unsuppress map in allowing specific prefix advertisement from aggregate
Difference is it can be configured on routers not originating the aggregate route
One map matches aggregate and router originating summary, other map matches specific prefix to advertise
Configuration:

ip prefix-list INJECT_PREFIX permit 10.0.1.0/24
ip prefix-list AGGREGATE permit 10.0.0.0/22
ip prefix-list ROUTE_SOURCE permit 155.1.37.3/32
!
route-map INJECT_MAP permit 10
 set ip address prefix-list INJECT_PREFIX
 set origin igp
!
route-map EXIST_MAP permit 10
 match ip address prefix-list AGGREGATE
 match ip route-source prefix-list ROUTE_SOURCE
! 
router bgp 300
 bgp inject-map INJECT_MAP exist-map EXIST_MAP

Route Filtering

Preferably done per-neighbor with route-maps, but can be applied directly with prefix list or ACL
neighbor 155.1.79.9 route-map FROM_R9 in for a route-map, with a deny in the route-map
neighbor 192.10.1.254 prefix-list BLOCK_222 in for prefix-list
neighbor 192.10.1.254 distribute-list BLOCK_222 in can be used for ACL filtering, not recommended

Maximum prefix

maximum-prefix can be used to limit allowed prefix # from a peer
By default shuts down connection, can use warning-only or restart <minutes> prefixes if desired
For example: neighbor 155.1.108.10 maximum-prefix 20 80 restart 3

BGP Dampening

Can be used to prevent table instability from flapping routes
Assigned to all prefixes with bgp dampening [<Half_Life> <ReuseLimit> <SuppressLimit> <MaximumSuppressTime>]
Can also be assigned to specific prefixes with route-maps:

ip as-path access-list 100 permit _100$
!
route-map DAMPENING
 match as-path 100
 set dampening 4 750 2000 16
!
router bgp 200
 bgp dampening route-map DAMPENING

BGP Outbound Route Filtering

Allows you to push a route filter to a remote neighbor
Benefit is decrease in unnecessary route information as filtering occurs on neighbor before prefixes sent
Both peers must have ORF capability enabled for it to function
Verified locally with: show ip bgp neighbors <ip>, remotely with: show ip bgp neighbors <ip> received prefix-filter

ip prefix-list ORF deny 112.0.0.0/8
ip prefix-list ORF permit 0.0.0.0/0 le 32
!
router bgp 100
 neighbor 155.1.45.5 capability orf prefix-list both
 neighbor 155.1.45.5 prefix-list ORF in

Redistribution

See Redistribution page for details on that

Useful debug/show commands

show ip bgp - Displays BGP table and attributes for each route
show ip bgp summary - Displays BGP neighbors, prefixes received from them, ASs, etc
show ip bgp neighbor <ip-address> [advertised-routes|received-routes] - See routes sent, received from peer
show ip protocols - Shows various information about active routing protocols
debug ip bgp - Debug all BGP address families, information
debug ip bgp updates - Debug BGP updates specifically

BGP Attributes

Weight

Only used on the router where it is configured
Used to affect outbound routing.
Higher value is better - default is 0 (32768 if originated locally)
Highest in BGP path selection algorithm
Cisco proprietary
Configuration:

route-map SET_WEIGHT
 match ip address|ip as-path ...
 set weight 100
!
router bgp 100
 neighbor <ip> route-map SET_WEIGHT in
 !
 neighbor <ip> weight <weight>

Local Preference

Locally significant (to the AS), does not transit outside AS.
Used to affect outbound routing.
Higher value is better - defaults to 100
Configuration:

route-map SET_LP
 match ip address|ip as-path ...
 set local-preference 1000
 !
router bgp 100
 neighbor <ip> route-map SET_LP in
 !
 bgp default local-preference <locpref>

AS-PATH

Changed for each AS prefix transits through
Used to affect inbound routing
Can be used to influence path selection with as-path prepending
Can be ignored with bgp bestpath as-path ignore - this is dangerous and not recommended
Can be limited in length with bgp maxas-limit <#>
Configuration:

route-map PREPEND
 match ip address|ip as-path ...
 set as-path prepend 100 100 100
    !
router bgp 100
 neighbor <ip> route-map PREPEND out

Origin Code

Rarely used to influence path selection - inflexible
IGP > EGP > Incomplete
Could affect inbound or outbound routing
Configuration:

route-map ORIGIN
 match ip address|ip as-path ...
 set origin [igp|egp|incomplete]
    
router bgp 100
 neighbor <ip> route-map ORIGIN out

MED

By default only compared when originating from same AS
Lowest is best - defaults to 0
Used to affect inbound routing
Can use bgp always-compare-med under router BGP to allow comparison of MED between different ASs
Configuration:

route-map MED
 match ip address|ip as-path ...
 set metric 1000
    
router bgp 100
 neighbor <ip> route-map MED out

Standard BGP Config

router bgp 100
 bgp router-id 4.4.4.4
 ! iBGP Neighbor
 neighbor 150.1.1.1 remote-as 100
 neighbor 150.1.1.1 update-source loopback0
 neighbor 155.1.1.1 route-reflector-client
 neighbor 155.1.1.1 next-hop-self # will only affect eBGP routes for RR clients
 ! eBGP Neighbor
 neighbor 150.1.3.3 remote-as 200
 neighbor 150.1.3.3 update-source loopback0
 neighbor 150.1.3.3 ebgp-multihop
 neighbor 150.1.3.3 send-community both
 neighbor 150.1.3.3 route-map FROM_R3 in
 neighbor 150.1.3.3 route-map TO_R3 in
 !
 network 155.1.4.0 255.255.255.0 route-map SET_COMMUNITY_50
 network 150.1.4.0 255.255.255.0
!
ip prefix-list BGP_PREPEND_1 seq 5 permit 150.1.4.0/24
!
ip prefix-list DEFAULT seq 5 permit 0.0.0.0/0
!
route-map FROM_R3 deny 10
 match ip address prefix-list DEFAULT
!
route-map FROM_R3 permit 10000
!
route-map TO_R3 permit 10
 match ip address prefix-list BGP_PREPEND_1
 set as-path prepend 100 100 100 100
!
route-map TO_R3 permit 10000
!
route-map SET_COMMUNITY_50 permit 10
 set community 50

Footnotes

BGP split horizon prevents loops in iBGP by disallowing advertisements received from one peer to be advertised on to another. This makes full mesh a requirement for iBGP unless using route reflectors and/or confederation↩︎