BGP Best-Path Selection

Now that you understand how BGP establishes sessions and exchanges routes, you must master something critical for the CCNP ENCOR exam:

How BGP decides which path to install when multiple paths exist.
BGP is a policy-based routing protocol.

That means:
You, as the network engineer, can influence path selection.

BGP Best-Path Algorithm

When a router receives multiple paths toward the same destination, it runs the BGP Best-Path Algorithm.
This Best-Path Algorithm allows us to influence the routing behavior of BGP.

We will now walk through the first four attributes in the exact order BGP evaluates them:

1. Weight

2. Local Preference

3. Locally Originated

4. AS-PATH

Until now, every prefix had a single path.

In real-world environments, this is rarely the case.
Routers often receive multiple paths to the same destination.

When that happens, BGP must decide:
Which path should be installed in the routing table?

Figure 2 – Two AS advertise prefix

In this topology:

• AS 200 advertises 209.165.200.224/27

• AS 300 advertises 209.165.200.224/27

• R1 receives both advertisements

You are on R1.
Your job is to determine which path will be selected, using the Weight attribute.

What is Weight?

Weight is a Cisco proprietary attribute that is local to the router. It is not advertised to neighbors and is the first attribute evaluated in the BGP decision process.

You can configure R1 to assign a weight to routes received from a specific neighbor. When that neighbor sends a prefix, the prefix arrives with the weight you defined.

Figure 3 – BGP Weight Attribute

The default values to remember are:

• 0 for routes learned from a neighbor

• 32768 for routes locally originated by the router

The Weight range goes from 0 to 65535.

Highest Weight wins.

Configuring Weight on R1

To configure Weight, you create a route-map that defines the action to apply.

Figure 4 – Setting BGP Weight

The route-map is named AS200, and its action is to set a weight value of 100.
Since there is no match clause, the route-map applies to all routes received from the targeted neighbor.

R1# conf t
Enter configuration commands, one per line.
R1(config)# route-map AS200 permit 10
R1(config-route-map)# set weight 100
R1(config-route-map)# exit

Then on R1, you enter the BGP process for AS 100 and apply this route-map inbound to the neighbor in AS 200.

R1(config)# router bgp 100
R1(config-router)# neighbor 10.0.11.2 route-map AS200 in
R1(config-router)# end

Verifying the Result

Now verify the BGP table:

R1# show bgp ipv4 unicast
BGP table version is 6, local router ID is 10.0.12.1
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,
              t secondary path,
Origin codes: i - IGP, e - EGP, ? - incomplete
RPKI validation codes: V valid, I invalid, N Not found

     Network              Next Hop        Metric LocPrf Weight Path
 *>  209.165.200.224/27   10.0.11.2            0           100 200 i
 *   209.165.200.224/27   10.0.12.2            0             0 300 i

The path through 10.0.11.2 now carries Weight 100, while the path through 10.0.12.2 still has Weight 0.

The best-path marker *> confirms R1 selected the AS 200 path to reach 209.165.200.224/27.

Figure 5 – Highest Weight Preferred

Because Weight is local to the router and never advertised, this decision applies only to R1.
No other router in the network is affected.

In the previous section, you saw that Weight influences path selection on a single router. However, in real-world networks, decisions often need to be consistent across the entire Autonomous System (AS).

This is where Local Preference comes into play.

Local Preference influences: Which exit point an entire AS should use to reach a destination.

Local Preference Topology

The scenario for this section is different from the Weight one.
You now focus on a network where AS 200 has two border routers, both connected to AS 300.

Figure 6 – Local preference topology

In this topology, AS 200 has two exit points toward AS 300:

• R1 connects to AS 300 through R3

• R2 connects to AS 300 through R4

• Both R1 and R2 receive the same prefix: 209.165.200.224/27

The question is:

Which exit point should AS 200 use?

Default Situation

By default, Local Preference equals 100. Verify what R2 sees first:

R2# show bgp ipv4 unicast
BGP table version is 9, local router ID is 10.0.24.1
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,
              t secondary path,
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 209.165.200.224/27   10.0.12.1            0    100      0 300 i
 *>  209.165.200.224/27   10.0.24.2            0    100      0 300 i

R2 sees two valid paths:

• One internal (iBGP) via R1, next-hop 10.0.12.1

• One external (eBGP) via R4, next-hop 10.0.24.2

Both paths have the same Weight (0) and the same Local Preference (100). Because all higher-priority attributes are equal, BGP keeps going down the decision list and eventually picks the eBGP path.

You will see why eBGP is preferred over iBGP in section 7. For now, just observe the result.

Figure 7 – R2 uses the eBGP path

R2 selects the direct eBGP path via 10.0.24.2.

What is Local Preference?

Local Preference is a well-known discretionary attribute that you use to influence outbound traffic from an AS. It has two key behaviors:

• It is propagated to all iBGP peers inside the AS

• It is never advertised to eBGP neighbors

Highest Local Preference wins.

The default value is 100.
Unlike Weight, Local Preference affects all routers inside the AS, not just the local one.

Configuring Local Preference

Now assume the routing policy is clear:

All traffic leaving AS 200 must exit through R1.

Figure 8 – Set default local preference

You configure Local Preference on R1 to make this happen.

R1# conf t
Enter configuration commands, one per line.
R1(config)# router bgp 200
R1(config-router)# bgp default local-preference 200
R1(config-router)# end

This command does two things:

• Every route R1 learns from its eBGP neighbor gets Local Preference 200

• R1 then advertises these routes to its iBGP peers (R2) with Local Preference 200

Figure 9 – Local preference 200 applied

Verifying the New Behavior

Now check R2 again:

R2# show bgp ipv4 unicast
BGP table version is 9, local router ID is 10.0.24.1
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,
              t secondary path,
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 209.165.200.224/27   10.0.12.1            0    200      0 300 i
 *   209.165.200.224/27   10.0.24.2            0    100      0 300 i

Three things changed:

• The path via R1 now carries Local Preference 200

• The path via R2 still carries the default 100

• The best-path marker *> moved to the iBGP path through R1

Figure 10 – Highest local preference wins

The path with Local Preference 200 wins, even though it is iBGP and the other one is eBGP.
This proves that Local Preference is evaluated before the eBGP vs iBGP rule.

The entire AS 200 now prefers exiting through R1 to reach 209.165.200.224/27.
The decision is consistent across all routers inside the AS.

After Weight and Local Preference, the next decision point in the BGP best-path algorithm is:

Prefer locally originated routes.

This rule applies when a router both owns a prefix (it has injected it into BGP itself) and receives the same prefix from a BGP neighbor. The router must decide which version to install.