During the build-up of SURFnet5, in the summer of 2001, IS-IS was used for IPv4 and IPv6. Since the transmission of SURFnet5, the 10G lambdas, is unprotected we used MPLS Traffic Engineering’s extension Fast Re-Route as the protocol to protect the network against failures by making available an alternative path well within the 50 msec time frame. However, MPLS-TE was only supporting IPv4, yielding a situation in which the IPv4 topology was different from the IPv6 topology. At that time, ISIS was not able to handle different topologies for the two protocols, and we had to move away from IS-IS for IPv6. We temporarily transitioned to RIPng for IPv6, while we stayed at IS-IS in combination with MPLS-TE FRR for IPv4.

In the mean time Cisco developed multi topology IS-IS, which enabled SURFnet to move away from RIPng back to the original plan. Before we went back to IS-IS for IPv6, we decided to abandon the MPLS-TE FRR ship for network management reasons, as we found the operations and maintaining of FRR too complex. At the same time, enhancements to IS-IS made it possible to fully rely on the routing protocols at the IP layer for the resilience in SURFnet5. During March 2004 the dual-stack approach was migrated to a 6PE implementation. The four core routers act as BGP route-reflectors in both the IPv4 as IPv6/6PE routing. All fifteen concentrator routers and border routers are BGP routereflector clients hanging of these BGP route-reflectors.