Configuring Load Balancing Across RSVP LSPs
By default, when you have configured several RSVP LSPs to the same egress router, the LSP with the lowest metric is selected and carries all traffic. If all of the LSPs have the same metric, one of the LSPs is selected at random and all traffic is forwarded over it.
Alternatively, you can load-balance traffic across all of the LSPs by enabling per-packet load balancing.
To enable per-packet load balancing on an ingress LSP, configure the policy-statement statement as follows:
You then need to apply this statement as an export policy to the forwarding table. For more information on how to configure the policy-statement statement, see the Routing Policy Configuration Guide.
Once per-packet load balancing is applied, traffic is distributed equally between the LSPs (by default).
You need to configure per-packet load balancing if you want to enable PFE fast reroute. To enable PFE fast reroute, include the policy-statement statement for per-packet load balancing shown in this section in the configuration of each of the routers where a reroute might take place. See also Configuring Fast Reroute.
You can also load-balance the traffic between the LSPs in proportion to the amount of bandwidth configured for each LSP. This capability can better distribute traffic in networks with asymmetric bandwidth capabilities across external links, since the configured bandwidth of an LSP typically reflects the traffic capacity of that LSP.
To configure RSVP LSP load balancing, include the load-balance statement with the bandwidth option:
You can configure this statement at the following hierarchy levels:
- [edit protocols rsvp]
- [edit logical-systems logical-system-name protocols rsvp]
Keep the following information in mind when you use the load-balance statement:
- If you configure the load-balance statement, the behavior of currently running LSPs is not altered. To force currently running LSPs to use the new behavior, you can issue a clear mpls lsp command.
- The load-balance statement only applies to ingress LSPs that have per-packet load balancing enabled.
- For Differentiated Services–aware traffic engineered LSPs, the bandwidth of an LSP is calculated by summing the bandwidth of all of the class types.
