Licensing Requirements and Limitations for Segment Routing
Involved Network Elements
SR-MPLS TE: When serving as a forwarder, a switch needs to work with the controller to provide the SR-MPLS TE function. SR-MPLS TE tunnels and tunnel attributes can be configured on the controller or forwarder. The controller calculates paths, generates and delivers labels, and maintains SR-MPLS TE tunnels.
SR-MPLS BE: Other network elements are required to support SR-MPLS BE.
Licensing Requirements
You can use the segment routing feature only after the MPLS function is enabled. The MPLS function is controlled by a license. By default, the MPLS function is disabled on a newly purchased switch. To use the MPLS function, apply for and purchase the license from the equipment supplier.
Version Requirements
Series |
Product Model |
Minimum Version Required |
|---|---|---|
CE12800 |
CE12804/CE12808/CE12812/CE12816/CE12804S/CE12808S |
V200R002C50 |
Feature Limitations
When deploying SR-MPLS TE on the switch, pay attention to the following points:
Among all VSs in port mode, SR-MPLS TE functions can only be configured in the admin-VS.
All VSs in group mode support SR-MPLS TE.
- When the controller is used to control SR-MPLS TE tunnel establishment between devices, do not configure link labels on the devices. Link labels may conflict with the configurations delivered by the controller, which may adversely affect the SR-MPLS TE service.
- If SR-MPLS TE functions have been configured on a switch, do not enable TRILL functions for interfaces on the SR-MPLS TE forwarding path. Otherwise, SR-MPLS TE functions do not take effect.
- When SR-MPLS TE is configured for path selection and equal-cost multi-path routing (ECMP) is configured for load balancing, paths in the output of the tracert lsp segment-routing te tunnel command may differ from actual ones.
- When the CE-L16CQ-FD, CE-L48XS-FD1, CE-L36CQ-FD1, CE-L36CQ-SD, CE-L48XS-FG, CE-L36CQ-FG and CE-L08CF-FG1 cards are configured to work in enhanced segment routing mode, you need to set the MPLS TTL processing mode to pipe on all SR nodes. Otherwise, the number of paths displayed after using the tracert function will decrease.
- When a switch works in non-enhanced mode in an IPv4 over SR-MPLS TE or IPv4 VPN over SR-MPLS TE scenario, ECMP and hot standby cannot be configured together. To configure both ECMP and hot standby, run the set forward capability enhanced command to set the interworking mode to enhanced mode.
- When a switch works in non-enhanced mode in an IPv4 over SR-MPLS TE group or IPv4 VPN over SR-MPLS TE group scenario, ECMP and hot standby cannot be configured together. To configure both ECMP and hot standby, run the set forward capability enhanced command to set the interworking mode to enhanced mode.
- If SR-LSP hot standby is configured on an SR-MPLS TE tunnel and a BFD for SR-MPLS TE tunnel session is configured, BFD for SR-MPLS TE LSP needs to be configured and the time for detecting the BFD for SR-MPLS TE tunnel needs to be longer than the time for detecting the BFD for SR-MPLS TE LSP. Otherwise, the BFD session may flap.
- If SR-LSP hot standby is configured on an SR-MPLS TE tunnel and an SBFD for SR-MPLS TE tunnel session is configured, SBFD for SR-MPLS TE LSP needs to be configured and the time for detecting the SBFD for SR-MPLS TE tunnel needs to be longer than the time for detecting the SBFD for SR-MPLS TE LSP. Otherwise, the SBFD session may flap.
- SBFD and TI-LFA can be configured only when the card interoperability mode is enhanced.
- In an SR-MPLS TE group scenario, the UCMP function cannot be configured for SR-MPLS TE tunnels.
- In an IPv4 VPN over SR-MPLS TE scenario, the UCMP function cannot be configured for SR-MPLS TE tunnels.
- When packets are redirected to SR-MPLS TE tunnels through MQC, the UCMP function cannot be configured for SR-MPLS TE tunnels.
- SBFD for SR-MPLS TE return packets are forwarded according to routes, rather than along the LSP specified by SR-MPLS TE. The forwarding of return packets depends on route reachability. In a route switchover scenario, the SBFD detection time must be longer than the route convergence time.
- After the anti-micro-loop function is enabled, the number of nodes on the TI-LFA backup path and the number of nodes on the primary path cannot exceed the number of label layers supported by the device.
- The SR-MPLS TE Group function does not take effect for IPv4 packets that are decapsulated through tunnels (such as VXLAN and GRE tunnels).
- When the MQC re-marked VPNv4 traffic enters an SR-MPLS TE group, the MQC-based traffic policy needs to be applied to the Layer 3 interface bound to a VRF.
- The traffic statistics collection period on an SR-MPLS TE tunnel must be larger than 30s. Otherwise, the statistics are inaccurate.
- If traffic statistics collection is enabled on an SR-MPLS TE tunnel interface and a VLANIF interface or a Layer 3 sub-interface on the tunnel, packets with the SR-MPLS TE tunnel interface as the next hop are counted as statistics on the SR-MPLS TE tunnel interface, but not statistics on the VLANIF interface or Layer 3 sub-interface.
- In an L3VPN over SR-MPLS TE scenario, if traffic statistics collection is enabled for both SR-MPLS TE tunnels and L3VPN, the function takes effect for SR-MPLS TE traffic but not for incoming L3VPN traffic.
- For cards except CE-L48XS-FDA, CE-L48XS-FD, CE-L48XS-FG, CE-L48XS-FD1, CE-L24LQ-FD, CE-L36LQ-FD, CE-L12CQ-FD, CE-L36CQ-FG, CE-L36CQ-FD1, CE-L36CQ-SD, CE-L16CQ-FD, CE-L08CF-FG1, and CE-L36CQ-FD, when multiple SR-MPLS TE tunnels load balance traffic using ECMP, traffic statistics collection in the inbound direction of an SR-MPLS TE tunnel becomes invalid.
When deploying SR-MPLS BE on the switch, pay attention to the following points:
Among all VSs in port mode, SR-MPLS BE functions can only be configured in the admin-VS.
All VSs in group mode support SR-MPLS BE.
When implicit-null label is enabled and LSPs whose next hop is label 3 or non-label 3 form a multi-path, Ps preferentially select the LSP with the smallest next-hop IP address among all non-outbound interfaces of the tunnel to forward traffic.
When implicit-null label is disabled in an SR-MPLS BE scenario, hash mis-sequencing may occur when MPLS packets are load balanced using ECMP and Eth-Trunk.
When MPLS packets carry four or more layers of labels (for example, in an IP over segment routing scenario), MPLS packets can be load balanced based on only the three outermost labels using ECMP and Eth-Trunk. In V200R019C00 and later versions, after the assign forward segment-routing hash enhanced command is run, Eth-Trunk or ECMP load balancing can be performed for packets with a maximum of five label stack layers.
The device does not support VPLS over SR-MPLS BE function.
- For a device at the penultimate hop, if the next hop is an LSP with implicit-null label 3 and an LSP with a non-3 label and the LSPs form an ECMP, only one LSP is selected to forward traffic, the ECMP load balancing function becomes invalid, and packets are lost during a link switchover. If the next hop is an LSP whose next hop is an implicit-null label 3 and an LSP with a non-3 label and the LSPs form a TI-LFA FRR, packet loss may occur when TI-LFA FRR switchback is performed. TI-LFA FRR switchover has the following impacts:
- When the primary LSP uses label 3 and traffic is forwarded through the backup LSP, the traffic is forwarded using the inner label after the outer label is popped out. In this case, the TTL of the inner label in the packet sent by the device is decreased by 1. As a result, the tracert path is incomplete.
- When the primary LSP uses a non-3 label and traffic is forwarded through the backup LSP, label 0 is added to the packet before the packet is forwarded.
- TI-LFA can be configured only when the card interoperability mode is enhanced.
- After the anti-micro-loop function is enabled, the number of nodes on the TI-LFA backup path and the number of nodes on the primary path cannot exceed the number of label layers supported by the device.