AS_Path

Four types of AS_Path attributes are available:

AS_Sequence: records in reverse order all the ASs through which a route passes from the local device to the destination.
AS_Set: records in random order all the ASs through which a route passes from the local device to the destination. The AS_Set attribute is used in route summarization scenarios.
AS_Confed_Sequence: records in reverse order all the sub-ASs within a BGP confederation through which a route passes from the local device to the destination.
AS_Confed_Set: records in random order all the sub-ASs within a BGP confederation through which a route passes from the local device to the destination.

Table 1 describes the AS_Path-based route selection rules.

**Table 1** AS_Path-based route selection rules
Item	Description
AS_Set	BGP takes an AS_Set as one AS number even if the AS_Set contains multiple AS numbers. When the aggregate (BGP) command is configured to generate a manually summarized route, if as-set is specified in the command, AS_Set will be carried in the summarized route. The information in the AS_Set is as follows: If the routes used in the summarization carry the same AS_Sequence, this AS_Sequence is carried in the summarized route, and the AS_Set of the summarized route is null. If the routes used in the summarization carry different AS_Sequences, all the AS numbers carried in the AS_Sequences are included in the AS_Set of the summarized route.
AS_Confed_Sequence and AS_Confed_Set	BGP ignores AS_Confed_Sequence and AS_Confed_Set when calculating the AS_Path length.
bestroute as-path-ignore	After the command is configured, BGP does not compare the AS_Path attribute during route selection.
apply as-path	The command can be run to configure an apply clause for a route-policy so that the ASs in the AS_Path of the route that matches the route-policy are cleared or replaced, or new ASs are added. NOTE: The configuration of the apply as-path command may change the traffic forwarding path, or cause routing loops or route selection errors. Therefore, exercise caution when configuring the command.
peer public-as-only	After the command is configured, BGP deletes private AS numbers (if any) from the AS_Path attribute before sending Update packets. Private AS numbers range from 64512 to 65534.
peer fake-as	After the command is configured, BGP can use a fake AS number to set up a BGP peer relationship. If the local device uses the actual AS number to establish an EBGP peer relationship with a remote device, the actual AS number is carried in the AS_Path of the route sent to the remote device. If the local device uses the fake AS number to establish the EBGP peer relationship, the fake AS number is carried in the AS_Path of the route sent to the remote device.
peer substitute-as	If the command is configured and an AS in the AS_Path carried in the route sent by a PE to the CE of the specified peer is the same as the AS of the CE, the PE replaces the AS with the local AS. NOTE: The peer substitute-as command applies only to PEs in BGP MPLS IP/VPN scenarios and may cause routing loops if it is improperly configured. Therefore, exercise caution when using the command.

During BGP route selection, BGP compares the AS_Path length by calculating the number of ASs included in the AS_Sequence if AS_Sequence is carried in a route. If both AS_Sequence and AS_Set are carried in the route, BGP considers the AS_Path length to be the number of ASs included in the AS_Sequence plus 1.

Deleting Private AS Numbers

As public AS resources are limited, carriers generally use private AS numbers when deploying VPNs. Private AS numbers, however, must not be advertised to the Internet because they may cause routing loops. In Figure 1, both ISP1 and ISP2 use 65001 as a private AS number.

Figure 1 Networking in which a private AS needs to be deleted

In Figure 1, Switch A advertises the route 10.0.0.0/8 to Switch D through ISP1 and ISP2. After receiving this route, Switch D checks its AS_Path. This AS_Path carries AS 65001, which is the same as the AS of Switch D. As a result, Switch D discards this route.

To address this problem, run the peer public-as-only command on Switch B so that Switch B deletes AS 65001 before adding AS 100 (public AS) to the AS_Path and forwarding the route to Switch C.

In the following situations, after the peer public-as-only command is configured, BGP does not delete any private AS number from the AS_Path:

The AS_Path of a route carries the AS number of the remote peer. In this case, deleting private AS numbers may lead to a routing loop.
The AS_Path carries both public and private AS numbers, which indicates that the route has passed through the public network. In this case, deleting private AS numbers may lead to incorrect traffic forwarding.

The preceding limitations also apply to confederation scenarios.

Adding AS Numbers

In Figure 2, AS 65005 imports three routes and advertises them to AS 65001 through two paths.

Figure 2 Networking in which new AS numbers are added to the AS_Path

Run the display bgp routing-table [ ip-address ] command to check the configurations.

# Display the routing table of Switch A.

[HUAWEIA] display bgp routing-table

 BGP Local router ID is 10.1.1.1
 Status codes: * - valid, > - best, d - damped,
               h - history,  i - internal, s - suppressed, S - Stale
               Origin : i - IGP, e - EGP, ? - incomplete


 Total Number of Routes: 6
      Network            NextHop        MED        LocPrf    PrefVal Path/Ogn

 *>   172.16.1.0/24      10.1.3.2                              0      65003 65005?
 *                       10.1.1.2                              0      65002 65004 65005?
 *>   172.16.2.0/24      10.1.3.2                              0      65003 65005?
 *                       10.1.1.2                              0      65002 65004 65005?
 *>   172.16.3.0/24      10.1.3.2                              0      65003 65005?
 *                       10.1.1.2                              0      65002 65004 65005?

[HUAWEIA] display bgp routing-table 172.16.1.0

 BGP local router ID : 10.1.1.1
 Local AS number : 65001
 Paths:   2 available, 1 best, 1 select
 BGP routing table entry information of 172.16.1.0/24:
 From: 10.1.3.2 (10.1.5.1)
 Route Duration: 00h00m56s
 Direct Out-interface: GigabitEthernet0/0/0
 Original nexthop: 10.1.3.2
 Qos information : 0x0
 AS-path 65003 65005, origin incomplete, pref-val 0, valid, external, best, select, active, pre 255
 Advertised to such 2 peers:
    10.1.3.2
    10.1.1.2

 BGP routing table entry information of 172.16.1.0/24:
 From: 10.1.1.2 (10.1.1.2)
 Route Duration: 00h34m43s
 Direct Out-interface: GigabitEthernet0/0/1
 Original nexthop: 10.1.1.2
 Qos information : 0x0
 AS-path 65002 65004 65005, origin incomplete, pref-val 0, valid, external, pre 255, not preferred for AS-Path
 Not advertised to any peer yet

The preceding command output shows that Switch A selects the route learned from Switch C because this route has a shorter AS_Path length than that learned from Switch B. To enable Switch A to select the route learned from Switch B, configure Switch B to reduce the AS_Path length of the route or configure Switch C to increase the AS_Path length of the route. In the following example, Switch C is configured to increase the AS_Path length of the route. The detailed configurations on Switch C are as follows:

#
bgp 65003
 #
 ipv4-family unicast
  undo synchronization
   peer 10.1.3.1 route-policy add_asn export            //Apply export policy named add_asn to routes to be advertised to 10.1.3.1.
  #
route-policy add_asn permit node 10                     //Define the first node of add_asn.
 if-match ip-prefix prefix1                             //Configure IP prefix list named prefix1.
 apply as-path 65003 65003 65003 additive               //Add 65003, 65003, 65003 to the AS_Path of the route that matches prefix1.
#
route-policy add_asn permit node 20                     //Define the second node of add_asn to permit all other routes.
#
ip ip-prefix prefix1 index 10 permit 172.16.1.0 24      //Define the first index of prefix1 to match the route 172.16.1.0/24.
ip ip-prefix prefix1 index 20 permit 172.16.2.0 24      //Define the second index of prefix1 to match the route 172.16.2.0/24.
ip ip-prefix prefix1 index 30 permit 172.16.3.0 24      //Define the third index of prefix1 to match the route 172.16.3.0/24.
#

Run the display bgp routing-table [ ip-address ] command to check the configurations.

# Display the routing table of Switch A.

[HUAWEIA] display bgp routing-table

 BGP Local router ID is 10.1.1.1
 Status codes: * - valid, > - best, d - damped,
               h - history,  i - internal, s - suppressed, S - Stale
               Origin : i - IGP, e - EGP, ? - incomplete


 Total Number of Routes: 6
      Network            NextHop        MED        LocPrf    PrefVal Path/Ogn

 *>   172.16.1.0/24      10.1.1.2                              0      65002 65004 65005?
 *                       10.1.3.2                              0      65003 65003 65003 65003 65005?
 *>   172.16.2.0/24      10.1.1.2                              0      65002 65004 65005?
 *                       10.1.3.2                              0      65003 65003 65003 65003 65005?
 *>   172.16.3.0/24      10.1.1.2                              0      65002 65004 65005?
 *                       10.1.3.2                              0      65003 65003 65003 65003 65005?

[HUAWEIA] display bgp routing-table 172.16.1.0

 BGP local router ID : 10.1.1.1
 Local AS number : 65001
 Paths:   2 available, 1 best, 1 select
 BGP routing table entry information of 172.16.1.0/24:
 From: 10.1.1.2 (10.1.1.2)
 Route Duration: 00h33m30s
 Direct Out-interface: GigabitEthernet0/0/1
 Original nexthop: 10.1.1.2
 Qos information : 0x0
 AS-path 65002 65004 65005, origin incomplete, pref-val 0, valid, external, best, select, active, pre 255
 Advertised to such 2 peers:
    10.1.3.2
    10.1.1.2

 BGP routing table entry information of 172.16.1.0/24:
 From: 10.1.3.2 (10.1.5.1)
 Route Duration: 00h02m12s
 Direct Out-interface: GigabitEthernet0/0/0
 Original nexthop: 10.1.3.2
 Qos information : 0x0
 AS-path 65003 65003 65003 65003 65005, origin incomplete, pref-val 0, valid, external, pre 255, not preferred for AS-Path
 Not advertised to any peer yet

The preceding command output shows that the AS_Path length of the route learned from Switch B is shorter than that of the route learned from Switch C and that the route learned from Switch B is selected as the optimal route. Table 2 shows the attribute comparison of the routes 172.16.1.0 learned from Switch B and Switch C.

**Table 2** Attribute comparison of the routes 172.16.1.0 learned from Switch B and Switch C
Route Attribute	Route Learned from Switch B	Route Learned from Switch C	Comparison
PrefVal	0	0	The same.
Local_Pref	-	-	The same.
Route type	Learned from a peer	Learned from a peer	The same.
AIGP	-	-	The same.
AS_Path	65002 65004 65005	65003 65003 65003 65003 65005	The route learned from Switch B is optimal.

AS numbers can be added to the AS_Path as required. However, if an AS number is added to the AS_Path of a route, the route cannot be received by devices in this AS. Therefore, the local AS number is added in most cases. For example in Figure 2, if Switch C adds AS 65001 to the AS_Path of a route before advertising the route to Switch A, Switch A will discard the route upon receipt because the route carries Switch A's AS number.

Replacing AS Numbers

When the apply as-path command is configured, if overwrite is specified in the command, the device will replace the AS numbers in the original AS_Path attribute to achieve the following goals:

Hide the actual path information.
Prevent a route from being discarded by replacing the AS_Path of the route with a shorter one if the as-path-limit command is configured on the device that receives this route.
Reduce the AS_Path length.

AS number replacement can also be used for the purpose of load balancing. For example in Figure 2, the apply as-path 65002 65004 65005 overwrite command can be configured on Switch A to replace the AS_Path of the route learned from Switch C so that the route has the same AS_Path as that of the route learned from Switch B, and the two routes are used to load-balance traffic. Detailed configurations on Switch A are as follows:

#
bgp 65001
 #
 ipv4-family unicast
  undo synchronization
   peer 10.1.3.2 route-policy replace_asn import        //Apply export policy named replace_asn to routes to be advertised to 10.1.3.1.
  #
route-policy replace_asn permit node 10                 //Define the first node of replace_asn.
 if-match as-path-filter filter1                        //Configure AS_Path filter named filter1.
 apply as-path 65002 65004 65005 overwrite              //Replace the AS_Path of the route that matches filter1 with 65002, 65004, 65005.
#
route-policy replace_asn permit node 20                 //Define the second node of replace_asn to permit all other routes.
#
ip as-path-filter filter1 permit ^65003                 //Define AS_Path filter named filter1 to match all the routes learned from AS 65003.
#

Run the display bgp routing-table [ ip-address ] command to check the configurations.

# Display the routing table of Switch A.

[HUAWEIA] display bgp routing-table

 BGP Local router ID is 10.1.1.1
 Status codes: * - valid, > - best, d - damped,
               h - history,  i - internal, s - suppressed, S - Stale
               Origin : i - IGP, e - EGP, ? - incomplete


 Total Number of Routes: 6
      Network            NextHop        MED        LocPrf    PrefVal Path/Ogn

 *>   172.16.1.0/24      10.1.1.2                              0      65002 65004 65005?
 *                       10.1.3.2                              0      65002 65004 65005?
 *>   172.16.2.0/24      10.1.1.2                              0      65002 65004 65005?
 *                       10.1.3.2                              0      65002 65004 65005?
 *>   172.16.3.0/24      10.1.1.2                              0      65002 65004 65005?
 *                       10.1.3.2                              0      65002 65004 65005?

The preceding command output shows that the AS_Path of the route received from AS 65003 has been replaced, after which the routes received from AS 65002 and AS 65003 have the same AS_Path. Run the maximum load-balancing 2 command on Switch A to set the maximum number of routes for load balancing to 2. Then, check the detailed route information. The route 172.16.1.0/24 is used in the following example:

[HUAWEIA] display bgp routing-table 172.16.1.0

 BGP local router ID : 10.1.1.1
 Local AS number : 65001
 Paths:   2 available, 1 best, 2 select
 BGP routing table entry information of 172.16.1.0/24:
 From: 10.1.1.2 (10.1.1.2)
 Route Duration: 19h57m51s
 Direct Out-interface: GigabitEthernet0/0/1
 Original nexthop: 10.1.1.2
 Qos information : 0x0
 AS-path 65002 65004 65005, origin incomplete, pref-val 0, valid, external, best, select, active, pre 255
 Advertised to such 2 peers:
    10.1.1.2
    10.1.3.2

 BGP routing table entry information of 172.16.1.0/24:
 From: 10.1.3.2 (10.1.5.1)
 Route Duration: 00h10m21s
 Direct Out-interface: GigabitEthernet0/0/0
 Original nexthop: 10.1.3.2
 Qos information : 0x0
 AS-path 65002 65004 65005, origin incomplete, pref-val 0, valid, external, select, active, pre 255, not preferred for router ID
 Not advertised to any peer yet

The preceding command output shows that the route learned from Switch B is optimal and is used along with the route learned from Switch C (not optimal) for load balancing. Check the information about the route 172.16.1.0/24 in the IP routing table.

[HUAWEIA] display ip routing-table

Route Flags: R - relay, D - download to fib, T - to vpn-instance
------------------------------------------------------------------------------
Routing Tables: Public
         Destinations : 9        Routes : 12

Destination/Mask    Proto   Pre  Cost      Flags NextHop         Interface

       10.1.1.0/30  Direct  0    0           D   10.1.1.1        GigabitEthernet0/0/1
       10.1.1.1/32  Direct  0    0           D   127.0.0.1       GigabitEthernet0/0/1
       10.1.3.0/30  Direct  0    0           D   10.1.3.1        GigabitEthernet0/0/0
       10.1.3.1/32  Direct  0    0           D   127.0.0.1       GigabitEthernet0/0/0
      127.0.0.0/8   Direct  0    0           D   127.0.0.1       InLoopBack0
      127.0.0.1/32  Direct  0    0           D   127.0.0.1       InLoopBack0
     172.16.1.0/24  EBGP    255  0           D   10.1.1.2        GigabitEthernet0/0/1
                    EBGP    255  0           D   10.1.3.2        GigabitEthernet0/0/0
     172.16.2.0/24  EBGP    255  0           D   10.1.1.2        GigabitEthernet0/0/1
                    EBGP    255  0           D   10.1.3.2        GigabitEthernet0/0/0
     172.16.3.0/24  EBGP    255  0           D   10.1.1.2        GigabitEthernet0/0/1
                    EBGP    255  0           D   10.1.3.2        GigabitEthernet0/0/0

The preceding command output shows that BGP has delivered the two routes with the same route prefix to the IP routing table for load balancing.

Clearing the AS_Path

When the apply as-path command is configured, if none overwrite is specified in the command, the device clears the AS_Path to hide the actual path information. If the AS_Path is null, BGP considers its length as 0 during route selection.