Over the years I’ve answered the question of, “Why MPLS?”, many times. There are many reasons from traffic engineering and unified provisioning/design to reducing route table size by implementing a BGP-free core. In this article we will review a routing issue and demonstrate how MPLS can be used to solve it by implementing a BGP-free core (and some other options).
Lets start with a basic routing issue demonstrated below:
In this example, R1 (126.96.36.199) and R2 (188.8.131.52) have an iBGP peering between them. The underlying network is running OSPF and the direct links and loopback0 interfaces are redistributed into OSPF. The problem in this case is that R1 learns the R2 networks via BGP (think of these as public IP prefixes) but the other routers in the network have no knowledge of these routes. When R1 tries to send traffic to the R2 networks the next hop routers drop the…
View original post 3,847 more words