Manipulate outgoing BGP Path selection.

[dschuckman]

Good Afternoon,

I have two Cisco 7206 VXR routers connected to two different upstream service providers using BGP. Both connections are OC3 and at this time we are utilizing 45% incoming traffic on both service providers, however we are using 30% outgoing on SP A and 0% on SP B. I was able to balance out the incoming through AS prepends etc... Anyways - this is not a major issue for me right now but my SP B just became multihomed with my SP A. So now when outgoing traffic selects its path to internet destination its sees 95% of the routes in the BGP topology table with one extra AS through SP B. I have talked to Cisco and they say I can't influence this I can only influence how I send my routes to my SP's. I just want to make sure that there is no way to accomplish this so that in the event that SP A ever becomes nearly saturated I can start influencing outbound traffic to take the less desirable AS path.

So the jist is how can I influence the routes that I am receiving to possibly remove the first AS or prepend an additional AS for incoming updates?

I know that I can select to have partial updates from SP if necessary or create fun route maps and specify certain networks to use one SP and others to use the second SP, but I just want to know if I can do something to the SP B as a hole to modify its number of AS hops and allow it to use some other metric to make decisions.

Thanks,

David

[Richard Burts]

David

It is true that you can not influence how many hops appear in advertisements from SP B. But that does not mean that you can not choose to prefer some routes from SP B. It is slightly complicated and will take some work on your part, but it is doable. By assigning a better local-preference to some routes to SP B, you can make those routes be preferred. So you would need to identify some routes (you probably do not want to prefer all routes from SP B) and give them the local pref. Or you might ask SP B to send you a subset of the complete routing table and a default route and you could local-preference for these.

So I do not think that you can do anything with SP B to reduce the number of hops. But you can use local preference to change the metric and make some different choices of outbound route.

HTH

Rick

[ilvasdetex]

1. Set Local Preference: Increase the local preference for a preferred path, influencing outgoing traffic decisions across your network.
2. Adjust MED: Use the Multi-Exit Discriminator (MED) to signal preferred exit points to neighboring ASes, though it's more commonly for incoming traffic, it can impact VAT-related cost optimizations.
3. AS-Path Prepending: Lengthen the AS path of less desirable routes to make them less attractive for outgoing traffic.
4. Route Filtering: Apply route filters to exclude certain paths from being considered for outbound traffic.
5. Next-Hop Self: Use the next-hop-self command on BGP neighbors to control the next-hop attribute and ensure consistent routing.
6. Community Attributes: Leverage BGP community attributes to tag and manipulate routes for specific outgoing path selection.
7. Weight Attribute: Use Cisco’s weight attribute like vat to prioritize specific routes for outgoing traffic, ensuring -efficient paths are chosen.
8. Policy-Based Routing (PBR): Implement PBR to force specific traffic flows through desired BGP paths.
9. Route Maps: Apply route maps to modify BGP attributes like local preference or MED to control outgoing path selection.
10. Conditional Advertisement: Advertise specific routes conditionally, based on the availability of preferred paths.
11. Aggregation: Aggregate routes to simplify routing tables, potentially influencing the outgoing path by reducing VAT-related routing costs.
12. BGP Peering: Adjust BGP peering sessions and configurations to prioritize certain routes for outgoing traffic, ensuring cost-effective VAT routing.

[paul driver]

Hello

---

@dschuckman wrote:

but I just want to know if I can do something to the SP B as a hole to modify its number of AS hops 

---

You could always tell your rtrs to ignore the AS-path comparison? meaning the as-path will be ignored from BGPs path selection process.

example:
RP/0/RP0/CPU0:ios#sh run router bgp
router bgp x
bgp bestpath as-path ignore