MLFR Bundle Data Plane

FRF.16.1 reuses the UNI/NNI fragmentation procedures defined in FRF.12. Frames on all FR SAP on the MLFR bundle have the UNI/NNI fragmentation header added regardless if they are fragmented or not. A separate sequence number state machine is used for each FR SAP configured on the bundle. The fragmentation threshold is configurable in the range 128 to 512 bytes.

To provide priority based scheduling of the FR SAP fragments over the bundle links, the user configures a FR scheduling class for each FR SAP configured on the bundle. As in MC-MLPPP, four scheduling classes are supported.

A separate fragmentation context is used by each FR SAP. FR SAPs of the same scheduling class share the same egress FR scheduling class queue with fragments of each SAP packets stored contiguously. The fragments from each scheduling class queue are then sprayed over the member links. Furthermore, the user may select the option to not fragment but spray the FR frames with the fragmentation header included over the member links.

Received fragments over the member links are re-assembled on a per SAP basis to re-create the original FR frame.

A user is not allowed to add an FR SAP with FRF.12 e2e fragmentation enabled to an MLFR bundle. Conversely, the user cannot enable FRF.12 e2e fragmentation on an FR SAP configured on an MLFR bundle. If an FR frame with the e2e fragmentation header is received on a bundle, it is forwarded if the FR SAP is part of an Fpipe service. It is discarded if the FR SAP is part of any other service.

Note that the operator must disable LMI before adding a link to an MLFR bundle. Also, the operator must shut down the bundle to change the value of the fragmentation threshold.

An FR SAP configured on an MLFR bundle can be part of a VLL, VPLS, IES, or VPRN service.