A few questions surrounding CEF responsibility…
When a traditional packet enters an LSR, an IP lookup is performed via the RIB and it’s forwarded via the FIB. When a label packet enters an LSR, it is switched via the LFIB and no lookup occurs. The router knows that it recieves a labeled packet or an IP Packet by looking at the protocol field in the Layer 2 header. Now, say a traditional packet enters an LSR and the LSR not only has the destination network in it’s RIB (and FIB) but is also running in the LIB and LFIB and can label switch via the LFIB. Which method does it choose if both services are available for forwarding? The book “MPLS Fundamentals” seems to outline a scenario that on how traffic gets forwarded depending on the service needed. If both provide a way to get there (which is the exact same path [IGP/EGP or LSP without MPLS TE]) how does it know to impose a label or just forward IP lookup? This decision has to be made somewhere upon traffic arriving at the ingress router.
This book also seems to use CEF to describe partial responsibility and other times, full responsibility in forwarding traffic for both traditional and label switched traffic. Does the CEF process also encapsulate the LFIB table and also control modification when the LIB updates itself from prefix changes? Certain parts of the book describe CEF only as being responsible on imposing/disposing LSRs and the LFIB for forwarding from label-to-label operations along with packet switching. When issued certain commands CEF seems to also hold valuable peices of information including labels for prefixes, so does it include label switching in it’s suit of switching methods?
I understand that the fundamental point the chapter is to differentiate between switching IP and switching Labels. The FIB switches IP, the LFIB switches labels. I would assume that CEF controls the LFIB since it’s documented that in order to fire up MPLS, CEF must be turned out. I’m open to suggestions and education.
Thanks,
Mike