CSPF Calculates a Traffic-Engineered Path

When the user configures a signaled Label Switched Path, the user specifies the address of the egress LER, as well as optional attributes, such as the LSPs priority and bandwidth requirements. The user can optionally specify a path of LSRs that the LSP must pass through on the way to the egress LER. When the user enables the signaled LSP, the Constrained Shortest Path First (CSPF) process on the ingress LER uses this information to calculate a traffic-engineered path between the ingress and egress LERs.

CSPF is an advanced form of the Shortest Path First (SPF) process used by IGP routing protocols. The CSPF process on the ingress LER uses the configured attributes of the LSP, user-specified path (when there is one), and the information in the Traffic Engineering Database (TED) to calculate the traffic-engineered path. This process consists of a sequential list of the physical interfaces that packets assigned to this LSP pass through to travel from the ingress LER to the egress LER. The traffic-engineered path takes into account the network topology, available resources, and user-specified constraints. The traffic-engineered path calculated by CSPF may or may not be the same as the shortest path that would normally be calculated by standard IGP routing protocols.

CSPF is enabled by default for signaled LSPs, but can be disabled. When signaled LSPs are configured without CSPF, the shortest path from the ingress LER to the egress LER is calculated using standard hop-by-hop routing methods. When the LSP also is configured to use a user-specified path, the device calculates the shortest path between each LSR in the path. As with CSPF, the output of this process is a fully specified path of physical interfaces on LSRs.

The advantage of configuring signaled LSPs without CSPF is that it can span multiple IS-IS levels. Since IS-IS LSPs with TE extensions have an area and level flooding scope, the information in an LSRs TED is relevant only to their area or level. Consequently, signaled LSPs that use CSPF can span only an IS-IS level. Signaled LSPs that do not use CSPF, because they do not rely on information in the TED, do not have this restriction.

Once the path for the LSP has been calculated, RSVP signaling then causes resources to be reserved and labels to be allocated on each LSR specified in the path. This may cause already existing, lower priority LSPs to be preempted. Once resources are reserved on all the LSRs in the path, the signaled LSP is considered to be activated; that is, packets can be forwarded over it.

The following sections provide additional information about the individual components of the process for activating traffic-engineered signaled LSPs, illustrated in Traffic-Engineered LSPs.