For IP/MPLS, a packet traversing the network stops at each LSR along the network path for processing to determine the next hop. The process first pops the ingress label from the label stack and queries it against the LFIB. The next-hop label found from the query pushes onto the top of the label stack, and the LSR sends the packet out to the associated next-hop interface. Each router along the service path performs the same function until the packet arrives at the destination. An analogy can explain the difference between a packet traversing an MPLS network versus an SR network.
Imagine driving to work in the morning. Every time you reach an intersection, you must provide a note to a traffic director to determine the next route. At the first intersection, the traffic director asks to see the note. With the appropriate information to look up the next route, the director gives you the next direction and another note to take to the next intersection. The journey continues; this action is repeated at every intersection until the destination. This process requires as many stops as there are intersections along the route. These multiple stops add time to the overall trip, and the longer the trip, the more time is spent exchanging information. This trip would be analogous to a journey through an IP/MPLS network.
Applying the same analogy to SR, consider the same trip to work through an SR network. Before leaving the point of origin, the initial note contains instructions on what direction to take at every intersection along the route. The result is no interactions with traffic directors, as the path is predetermined. This approach saves time and reduces overhead. The behavior is analogous to the implementation of the SR label stack.
As mentioned previously, the SR protocol has determined the path through the network upon entry to an SR network. An ordered list of instructions, in the form of SIDs, is calculated by the IGP, encoded as labels, and pushed onto the packet, guiding the network traffic. The network administrator can also choose to engineer the path using SR policies. These policies might diverge from the shortest path calculation determined by the IGP to recover from link failures or bypass a specific link. Additionally, SR policies may trigger changes in network paths by dynamic or static conditions in the network.
In a standard IGP network, reconvergence of a network after a link failure could take from hundreds of milliseconds to tens of seconds. This time frame for recovery is not acceptable to mission-critical utility application traffic. SR networks might utilize topology-independent loop-free alternate (TI-LFA) to improve the resiliency and reliability of the operations network. TI-LFA reduces the reconvergence time to tens of milliseconds by precalculating alternate next-hops if the active next-hop path fails. This recovery time closely mirrors capabilities provided by RSVP FRR in traditional MPLS networks.
As utility operations and network requirements continue to evolve, new protocols like SR improve networks by addressing growing challenges of more traffic with better convergence and sustained resiliency. SR can implement networks in place of LDP and RSVP for label distribution to provide ease of use and feature-rich policies requiring less manual configuration and overhead to create those configurations.