Guidelines for PCC-Initiated and PCC-Controlled LSPs

The router supports both a full CSPF and a basic hop-to-label translation path computation methods for a SR-TE LSP. In addition, the user can configure a path for the SR-TE LSP by explicitly entering SID label values.

The ingress LER has a few ways to detect a path is down or is not optimal and take immediate action:

With both the hop-to-label path computation method and the user configured SID labels, the ingress LER does not monitor network events which affect the reachability of the adjacency SID or node SID used in the label stack of the LSP, except for the top SID. As a result, the label stack may not be updated to reflect changes in the path except when seamless BFD is used to detect failure of the path. It is therefore recommended to use this type of SR-TE LSP in the following configurations only:

The user can also configure a SR-TE LSP with a single loose-hop using the anycast SID concept to provide LSR node protection within a given plane of the network TE topology. This is illustrated in Figure 1. The user configures all LSRs in a given plane with the same loopback interface address, which must be different from that of the system interface and the router-id of the router, and assigns them the same node-SID index value. All routers must use the same SRGB.

Figure 1. Multi-plane TE with Node Protection

Then user configures in a LER a SR-TE LSP to some destination and adds to its path either a loose-hop matching the anycast loopback address or the explicit label value of the anycast SID. The SR-TE LSP to any destination will hop over the closest of the LSRs owning the anycast SID because the resolution of the node-SID for that anycast loopback address uses the closest router. When that router fails, the resolution is updated to the next closest router owning the anycast SID without changing the label stack of the SR-TE LSP.