BGP Path Attribute Manipulation

1. Inbound vs. Outbound Traffic Engineering

Before diving into configuration, it's crucial to distinguish between the two primary goals of BGP traffic engineering:

• Controlling Outbound Traffic (Egress): This involves choosing which exit point your Autonomous System (AS) uses to send traffic to an external destination. Because you control all the routers within your AS, you have authoritative control over this decision. The primary attribute for this is Local Preference.

• Influencing Inbound Traffic (Ingress): This involves influencing which of your entry points other ASes use to send traffic to your network. You don't control your provider's routers, so you can't command them. Instead, you advertise your routes in a way that makes one path appear more or less desirable. The most common tool for this is AS-path prepending.

We will tackle each of these in turn.

2. Controlling Outbound Traffic with Local Preference

Recall from our last lesson that Local Preference is evaluated before AS-path length, and the highest value wins. It is propagated to all iBGP peers within an AS, making it the perfect tool to ensure all routers in your network agree on the best way to exit. The default Local Preference is 100.

Let's imagine your enterprise network (AS 100) has two connections to the internet via two different ISPs (AS 200 and AS 300). You want all traffic destined for a partner network (e.g., 8.8.8.8/32) to use the link to AS 300, perhaps because it offers a better service level agreement (SLA).

To achieve this, you need to set a higher Local Preference on the route to 8.8.8.8/32 as it is learned from AS 300. This is done using a route map.

The following video provides a concise and clear demonstration of this entire process.

Configuration Breakdown

As you saw in the video, the configuration process follows these key steps:

1. Identify Traffic: Define the routes you want to influence. This is typically done with a prefix list.

   ip prefix-list PL_TARGET_PREFIX permit 8.8.8.8/32
   

2. Create a Route Map: The route map matches the traffic and sets the attribute. Remember to include a second, empty permit statement to avoid implicitly denying all other routes.

   route-map RM_SET_LP permit 10
    match ip address prefix-list PL_TARGET_PREFIX
    set local-preference 200
   !
   route-map RM_SET_LP permit 20
   

3. Apply the Route Map: You apply this policy to the eBGP neighbor from which you want to prefer the routes. A crucial point here is the direction: you apply it inbound.

   router bgp 100
    neighbor 13.0.0.3 remote-as 300
    neighbor 13.0.0.3 route-map RM_SET_LP in
   

Why apply the route map in?
You are modifying the attribute of a route as it is being learned and entering your AS. By setting the Local Preference at the ingress edge router, that router will then advertise the route to its iBGP peers with the new, higher Local Preference (200), ensuring the entire AS prefers this path.

3. Influencing Inbound Traffic with AS-Path Prepending

Now for the more subtle art of influencing your neighbors. Since you can't configure their routers, you make one of your paths look less attractive. The most common way to do this is to make the AS-path seem longer. This technique is called AS-path prepending.

You'll advertise a route to your non-preferred ISP with your own AS number repeated several times. Since BGP's default behavior is to choose the path with the shortest AS-path, other routers on the internet will see the prepended path as longer and less desirable.

This next video is an excellent, practical demonstration from a real-world perspective. The presenter's goal is to shift traffic away from a more expensive provider link, a scenario you would absolutely encounter as a solutions engineer.

Configuration Breakdown

The configuration for AS-path prepending is conceptually similar to what we did for Local Preference but with two key differences: the set command used and the direction of the route map.

1. Identify Traffic: Use a prefix list to match the networks you are advertising.

   ip prefix-list PL_MY_PREFIXES permit 63.232.144.0/24
   

2. Create a Route Map: Match the prefix list and use the set as-path prepend command. You repeat your own AS number. Adding it 3 times makes the path 3 "hops" longer.

   route-map RM_AS_PREPEND permit 10
    match ip address prefix-list PL_MY_PREFIXES
    set as-path prepend 19468 19468 19468
   !
   route-map RM_AS_PREPEND permit 20
   

3. Apply the Route Map: Apply the policy to the eBGP neighbor towards which you want to make the path less preferable. The direction is outbound.

   router bgp 19468
    neighbor 72.44.x.x remote-as 209
    neighbor 72.44.x.x route-map RM_AS_PREPEND out
   

Why apply the route map out?
You are modifying the attribute of a route as you are advertising it to your external neighbor. You are altering the information you send them to influence their decision.

A key takeaway from the video is that this is not a guaranteed method. A neighboring AS might have its own policies (like a higher Local Preference) that override your attempt to influence them. However, for most default scenarios, it is a very effective technique.

4. A Brief Look at Configuring Weight and MED

To complete the picture, let's briefly cover how to configure two other attributes we've discussed: Weight and MED. While Local Preference and AS-path prepending are the most common tools for traffic engineering, it's good to know how to set these as well.

The following article provides detailed configuration examples for all major BGP attributes.

Here's a quick summary of the configuration:

• Weight:

  • Goal: To influence path selection on a single router. Highest weight wins.
  • How:
    • Simple method: neighbor <ip> weight <value> under router bgp.
    • Flexible method: set weight <value> in a route-map applied inbound.

• MED (Multi-Exit Discriminator):

  • Goal: To influence how a single neighboring AS chooses to enter your network (if you have multiple links to that same AS). Lowest MED wins.
  • How: set metric <value> in a route-map applied outbound.

Conclusion

In this lesson, you've made the leap from analyzing BGP paths to actively controlling them. You now have the fundamental tools of BGP traffic engineering at your disposal, a skill set that is essential for designing robust and efficient enterprise networks.

Key Takeaways:

• You can control outbound (egress) traffic for your entire AS using Local Preference. Higher is better, and the policy is applied inbound on your edge router.
• You can influence inbound (ingress) traffic from external networks using AS-path prepending. Shorter is better, so you lengthen the path on non-preferred links. The policy is applied outbound.
• Route maps are the essential mechanism for implementing BGP policies, using match criteria to identify routes and set commands to modify attributes.
• Always remember to add a permit statement at the end of your route map to prevent unintended filtering of other routes.

Preview of the Next Lesson:

Now that you have a solid grasp of how to manage routing policy within your AS, a new challenge arises: as your network grows, how do you manage the full mesh of iBGP sessions required to propagate these policies? In the next lesson, we will solve this problem by learning how to configure and verify iBGP route reflection for scalability.