SR-MPLS with IPv6 Control Plane (Flex-Algo)

As IPv4 addresses are depleting fast, IPv6 is gaining more and more popularity, so bringing up SR‐MPLS in an IPv6‐only control plane is becoming a norm. As discussed earlier, putting everything in IPv6 lowers the administrative burden and the dual‐stack complexity. We employed two core spine routers in the center and supported more than one PE node interconnected by all with IPv6 loopbacks and link addresses. In this setup, IS‐IS served as the IGP, advertising the Node SIDs, Adjacency SIDs, and three Flex‐Algo SIDs, which specifically target the following: delay metric (FA128), traffic‐engineering metric (FA129), and admin group exclusions (FA130).
Instead of utilizing probe-based mechanisms, we used static or IGP‐learned metrics to choose paths. Whenever link metrics were updated (for example, increasing the delay or adjusting an exclude flag), each instance of Flex‐Algo rerouted traffic seamlessly without even a single drop of packets. The control plane was, however, intended for IPv6 uses only, but this did not prevent the system from supporting both IPv4 and IPv6 dual-stack services. BGP color communities enabled us to assign different color communities to different prefixes in the same VRF; we managed to influence how the traffic is routed per prefix or per group of prefixes.
Executing the correct Flex algorithms for a particular service was accomplished by simply tagging every service to a distinct color rather than by a sequence of manual command executions. In general, it showed that SR-MPLS in an IPv6-only control, along with the Flex-Algo constraints, is supporting IPv4 and IPv6 services while keeping the network simple and flexible.