The following are the limitations of the CBF feature.
CBF applies to packets of IPv4 and IPv6 BGP prefixes only. CBF does not apply to IGP prefixes and static route prefixes resolved over IGP IPv4 shortcuts. The latter are forwarded using regular ECMP over the entire set of up to 64 tunnel next-hops.
CPM originated packets on the router, including the control plane and OAM packets, are forwarded over a single LSP from the set of LSPs the packet's FC is mapped to, as per the CBF configuration. The CPM, however, only maintains a maximum of 64 next-hops for a specified destination prefix. Therefore, if there are multiple BGP next-hops for a prefix, CPM selects 64 tunnel next-hops by cycling over the BGP next-hops in ascending order. Then, the first LSP in the first set ID that the FC of the packet maps to is selected to forward the packet.
Furthermore, the CBF information consistency check, the CBF default set determination, and the CBF set failover procedures are applied to this set of 64 tunnel next-hops.
The user can configure the SGT-QoS feature to change the DSCP and FC of CPM-originated packets of a specific control plane protocol to select an LSP from a different set ID. This configuration allows, for instance, the forwarding of BGP Keep-Alive packets over an LSP of the same set ID as that of the data plane packets of the BGP prefixes destined for the same BGP next-hop.
Weighted ECMP, at the transport tunnel level of BGP prefixes over IGP shortcuts, and the CBF feature on a per-BGP next-hop basis are mutually exclusive. Specifically, if the user enables both weighted ECMP (config>router>weighted-ecmp) and CBF (config>router> class-forwarding), weighted ECMP applies as long as all the LSPs used as tunnel next-hops to reach the BGP next-hop of a prefix have a user-configured weight. Otherwise, the CBF feature applies as per the procedures described in Feature behavior.