BGP AS-Path Prepending: Technical Details

Monday, March 2, 2009 06:47 CET… updated on Monday, December 7, 2020 11:12 UTC

BGP AS-Path Prepending: Technical Details

I thought I knew all there is to know about the AS-path prepending before the February 2009 incident, which prompted me to focus on this particular Cisco IOS feature.

For example, did you know you could do inbound AS-path prepending? I didn’t, until Rodney Dunn from Cisco mentioned it in an e-mail exchange. Did you ever wonder whether the AS-path prepending affects inbound or outbound AS-path filters? I had a hunch it doesn’t, but was never sure. Time to figure out all the gory details…

The following text written by Ivan Pepelnjak in 2009 was originally published on CT3 wiki. That web site became unreachable in early 2019. We retrieved the original text from the Internet Archive, cleaned it up, updated it with recent information if necessary, and republished it on ipSpace.net blog on December 7, 2020

AS Path Prepending 101

AS-path prepending is the manipulation of the BGP AS-path attribute beyond the insertion of local AS number on outgoing EBGP updates. Extra AS-numbers are inserted (prepended) at the beginning of AS-path, just after the local AS-number.

Cisco IOS supports inbound and outbound AS-path prepending on EBGP sessions. AS-path prepending does not work on IBGP sessions.

Outbound AS-path prepending can be used as the last-resort mechanism to influence global BGP routing policies in BGP multi-homing scenarios where all other methods (Multi-exit Discriminator or Local preference manipulation through BGP communities) don’t work due to lack of upstream ISP’s support or due to the wide difference in upstream ISP’s connectivity to the internet core.

Only a few copies of the local AS-number should be prepended to the AS-path when you use AS-path prepending to influence the BGP routing policies. If your problem cannot be solved by prepending less than 10 copies of the local AS-number, you should use other mechanisms (see also the prepending statistics by Geoff Huston).

You should not prepend long AS-path segments, modify inbound updates with inbound AS-path prepending or prepend non-local AS numbers outside of a lab environment. Never prepend long AS-path segments in the Internet.

Test Network Diagram

Throughout the article, the following test network will be used to generate router printouts. The initial router configurations are summarized in the Initial router configuration section. The routers were running Cisco IOS release 15.6(1)T.

Cisco IOS Configuration

BGP AS-path prepending is configured with the set as-path prepend statement within a route-map. The route-map can then be applied to inbound or outbound updates received or sent to an EBGP peer.

The AS-path prepending does not work on IBGP sessions or when the route-map is used in a network statement. In both cases, the set as-path prepend route map command is ignored without an error message.

Outbound AS-Path Prepending

For example, to prepend two extra copies of the local AS-number to the outbound BGP updates sent from E1 to R1, you could use the following configuration on E1:

Outbound AS-path prepending on E1

router bgp 65000
 bgp log-neighbor-changes
 network 192.168.0.1 mask 255.255.255.255
 neighbor 10.0.0.2 remote-as 64800
 neighbor 10.0.0.2 description BGP sesssion to r1
 neighbor 10.0.0.2 next-hop-self
 neighbor 10.0.0.2 route-map prepend out
!
route-map prepend permit 10
 set as-path prepend 65000 65000

The AS-path prepending is applied to EBGP updates when they are sent or received. A change in the route-map configuration or BGP routing protocol configuration does not generate the outbound BGP updates, although they might be generated at the next BGP scanner run (default value: 60 seconds). To force the change in the BGP routing policy after a route-map has been updated, use the clear ip bgp neighbor soft out command.

Old Cisco IOS releases never sent BGP updates to BGP neighbors after a change in an outbound route-map. IOS release 15.6(1)T consistently generated an outbound update at the next BGP generic scan after a route map changed.

After the changed BGP updates have been received by R1, its BGP table reflects the modified AS-path sent by E1. Due to longer AS-path on prefixes received over the E1-R1 link, R1 prefers IBGP paths received from R2; you’ve turned the link between E1 and R1 into a backup link.

r1#show ip bgp
BGP table version is 9, local router ID is 192.168.0.2
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
              r RIB-failure, S Stale, m multipath, b backup-path, f RT-Filter,
              x best-external, a additional-path, c RIB-compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found

     Network          Next Hop            Metric LocPrf Weight Path
 *>i 192.168.0.1/32   10.0.0.3                 0    100      0 65000 i
 *                    10.0.0.1                 0             0 65000 65000 65000 i
 *>  192.168.0.2/32   0.0.0.0                  0         32768 i
 *>i 192.168.0.3/32   10.0.0.3                 0    100      0 i

Inbound AS-Path Prepending

You can use set as-path prepend command in an inbound route map. For example, to add an extra copy of the remote AS number to inbound updates received from E1, use the following configuration on R2:

router bgp 64800
 bgp log-neighbor-changes
 network 192.168.0.3 mask 255.255.255.255
 neighbor 10.0.0.1 remote-as 65000
 neighbor 10.0.0.1 description BGP sesssion to e1
 neighbor 10.0.0.1 next-hop-self
 neighbor 10.0.0.1 route-map prependIn in
!
route-map prependIn permit 10
 set as-path prepend last-as 1

The set as-path prepend last-as prepends the first AS in the AS-path (neighbor’s AS). You can use this command in an inbound route-map to simplify your configuration and make the route-map independent of the neighbor’s AS number.

Do not use set as-path prepend last-as in an outbound route-map. The set statement would be executed before the local AS number is prepended to the AS-path, resulting in an “interesting” behavior.

After R2 receives new EBGP update from E1 (for example, triggered by the clear ip bgp neighbor soft in command or by a reset of the BGP session), the AS-path stored in the BGP table on R2 contains two copies of AS 65000:

r2#sh ip bgp
BGP table version is 10, local router ID is 192.168.0.3
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
              r RIB-failure, S Stale, m multipath, b backup-path, f RT-Filter,
              x best-external, a additional-path, c RIB-compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found

     Network          Next Hop            Metric LocPrf Weight Path
 *>  192.168.0.1/32   10.0.0.1                 0             0 65000 65000 i
 *>i 192.168.0.2/32   10.0.0.2                 0    100      0 i
 *>  192.168.0.3/32   0.0.0.0                  0         32768 i

Troubleshooting AS-path Prepending

Outbound AS-path prepending can be reliably observed only in the BGP table of the EBGP peers. The show ip bgp neighbor address advertised-routes command does not display the results of the outbound route-map; it displays the routes in the local BGP table that are advertised to the specified neighbor.

For example, the show ip bgp neighbor 10.0.0.2 advertised command executed on E1 displays empty AS-path even though two copies of AS65000 are prepended to the AS-path in the outgoing EBGP updates.

e1#sh ip bgp nei 10.0.0.2 advertised-routes
BGP table version is 6, local router ID is 192.168.0.1
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
              r RIB-failure, S Stale, m multipath, b backup-path, f RT-Filter,
              x best-external, a additional-path, c RIB-compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found

     Network          Next Hop            Metric LocPrf Weight Path
 *>  192.168.0.1/32   0.0.0.0                  0         32768 i

The debug ip bgp updates command is also useless, as it displays the contents of the local BGP table, not the output of the outbound route-map.

AS-path Prepending and AS-path Filters

The prepended portion of the AS-path is not used in inbound or outbound AS-path filters; either in the match as-path conditions or in the per-neighbor filter lists. The AS-path filters always match the actual AS-path received from the BGP neighbor (when applied in the inbound direction) or the AS-path of the entry in the BGP table (when applied in the outbound direction).

Outbound Prepending and AS-path Tests within the Route Map

The AS-path prepending does not influence the AS-path filters used within the route-map in the match as-path command. These filters are always matched against the original BGP entry in the BGP table. AS numbers specified with the set as-path prepend commands are accumulated and prepended to the AS-path attribute after the route map processing is completed.

To test the IOS behavior, we’ll modify the outbound prepend route map on E1 to include an AS-path filter. The continue 20 statement in the first part of the route map ensures that the whole route map is executed.

Testing whether route map match statements use prepended AS-path

ip as-path access-list 100 permit ^$
!
route-map prepend permit 10
 set as-path prepend 10
 continue 20
!
route-map prepend permit 20
 match as-path 100
 set as-path prepend 20

If the AS-path filters would match the prepended AS-path, the match as-path 100 test in the route-map prepend permit 20 statement would not succeed and only a single AS-number would be prepended to the AS-path. However, the display of BGP table on R1 verifies that E1 prepended both AS-numbers to the AS-path:

r1#sh ip bgp
BGP table version is 10, local router ID is 192.168.0.2
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
              r RIB-failure, S Stale, m multipath, b backup-path, f RT-Filter,
              x best-external, a additional-path, c RIB-compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found

     Network          Next Hop            Metric LocPrf Weight Path
 *>i 192.168.0.1/32   10.0.0.3                 0    100      0 65000 65000 i
 *                    10.0.0.1                 0             0 65000 20 10 i
 *>  192.168.0.2/32   0.0.0.0                  0         32768 i
 *>i 192.168.0.3/32   10.0.0.3                 0    100      0 i

Outbound Prepending and Outbound Neighbor Filter Lists

The outbound AS-path prepending does affect the operation of outbound filter lists. The AS-path access list specified in the filter-list out BGP neighbor option matches AS-paths in the BGP table, not the prepended paths generated by outbound route map.

If we modify the BGP router configuration on E1 to include an outbound filter-list on neighbor 10.0.1.6, E1 still sends the same prefix to R1, proving that the outbound filter-list does not test the prepended path.

Combination of AS-path prepending and outbound filter-list on E1

router bgp 65000
 bgp log-neighbor-changes
 network 192.168.0.1 mask 255.255.255.255
 neighbor 10.0.0.2 remote-as 64800
 neighbor 10.0.0.2 description BGP sesssion to r1
 neighbor 10.0.0.2 next-hop-self
 neighbor 10.0.0.2 route-map prepend out
 neighbor 10.0.0.2 filter-list 100 out
!
ip as-path access-list 100 permit ^$
!
route-map prepend permit 10
 set as-path prepend 65000 65000 65000

Outbound updates are still sent to R2

e1#debug ip bgp updates 10.0.0.2 out
BGP updates debugging is on for neighbor 10.0.0.2 (outbound) ↲ 
for address family: IPv4 Unicast
e1#clear ip bgp * soft out
e1#
BGP(0): 10.0.0.2 NEXT_HOP is set to self for net 192.168.0.1/32,
BGP(0): (base) 10.0.0.2 send UPDATE (format) 192.168.0.1/32, ↲
next 10.0.0.1, metric 0, path Local

Inbound Prepending and Inbound Neighbor Filter Lists

The inbound AS-path filter-list is applied before the inbound route-map. The AS-path attribute tested by the inbound AS-path access-list is thus the original AS-path sent by the EBGP neighbor, not the AS-path modified by inbound prepending.

You can use modified configuration of R2 to test this behavior.

router bgp 64800
 bgp log-neighbor-changes
 network 192.168.0.3 mask 255.255.255.255
 neighbor 10.0.0.1 remote-as 65000
 neighbor 10.0.0.1 description BGP sesssion to e1
 neighbor 10.0.0.1 next-hop-self
 neighbor 10.0.0.1 route-map prependIn in
 neighbor 10.0.0.1 filter-list 100 in
!
ip as-path access-list 100 permit ^65000$
!
route-map prependIn permit 10
 set as-path prepend last-as 2

The inbound filter-list matches the expected AS-path (65000) and the BGP debugging confirms the inbound EBGP update is accepted.

r2#debug ip bgp upd 10.0.0.1 in
BGP updates debugging is on for neighbor 10.0.0.1 (inbound) ↲
for address family: IPv4 Unicast
r2#clear ip bgp * soft in
r2#
...
BGP(0): 10.0.0.1 rcvd UPDATE w/ attr: nexthop 10.0.0.1, ↲
origin i, metric 0, merged path 65000, AS_PATH

The AS-path in the BGP table on R1 contains the received AS-path as well as the results of the inbound prepending:

r2#show ip bgp
BGP table version is 10, local router ID is 192.168.0.3
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal,
              r RIB-failure, S Stale, m multipath, b backup-path, f RT-Filter,
              x best-external, a additional-path, c RIB-compressed,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found

     Network          Next Hop            Metric LocPrf Weight Path
 * i 192.168.0.1/32   10.0.0.2                 0    100      0 65000 20 i
 *>                   10.0.0.1                 0             0 65000 65000 i
 *>i 192.168.0.2/32   10.0.0.2                 0    100      0 i
 *>  192.168.0.3/32   0.0.0.0                  0         32768 i