default routes scenario

[alecardo]

Hi

Coming from a (mostly) switching and data center background, I'm still wrapping my head around some core routing concepts. 

Take two routers, R1 and R2, sharing a BGP/OSPF/EIGRP adjacency. When and why would you

a) *manually* configure a default route in R1 pointing to R2? or

b) advertise the default route from R2 to R1? (ex: "neighbor x.x.x.x default-originate" in BGP)

Hope you can provide some insight. Thank you!

[Meheretab Mengistu]

Hi,

Take two routers, R1 and R2, sharing a BGP/OSPF/EIGRP adjacency. When and why would you
a) *manually* configure a default route in R1 pointing to R2? or

If the only upstream connection you are having is through R2, you will configure "default route" manually so that you will lower or minimize the load over CPU, and minimize re-calculations when there is a change in the topologies.

b) advertise the default route from R2 to R1? (ex: "neighbor x.x.x.x default-originate" in BGP)

 You will advertise default route if R1 is the only ISP connection. The reason is that whether you provide default route, partial or full BGP paths, all traffic from R2 comes through R1. As a result, it does not make sense to advertise differently. Instead, advertising default route will save a memory, and CPU load for R2. 

HTH,

Meheretab

[alecardo]

Thanks Meheretab, although I still don't see an advantage of why using one or another in the scenario I'm studying:

[campus L3 core]  >  R2  >  [ISP]  <  R1

How would my campus L3 core benefit from either scenario? Pros/Cons of manually configuring a default route from R2 to R1? Pros/Cons of R1 advertising a default route to R2?

Thanks!

[chrihussey · Accepted solution]

CPU considerations aside, with a static default route it will be active as long as the interface is up. If the interface goes down then it would be nullified or no longer be in the routing table. The disadvantage here is that if the interface is up and the neighbor of whatever protocol was down the route would still be active and traffic could get black holed. This would be especially true in a situation where one is routing to a firewall and does not know the status of the peer on the other side.

By getting the default route advertised by the peer guarantees the routing at the other end of the link is active. However, short of the interface going down, there may be a slight delay before the router is aware the neighbor is down. Regardless, you avoid the first scenario.

Finally, it all depends on the network topology and what one needs to achieve. With one point of exit or a direct connection, a static may be just fine. If trying to achieve redundancy with multiple points of exit or indirect connections advertisement using a routing protocol may be the way to go.

Hope this helps

[alecardo]

Thank you very much! That makes it a lot more clear. I agree, in the end it depends on the final solution, but you answer helped identify the pros and cons of each option.