Why CEF needed in MPLS Network??

Question

I have read the MPLS Fundamentals book by Luc De Ghein, So I understand from the from book that cef needs to enabled in edge routers to tag or untag labels (for Ip packets). I am eager to know why Mpls (Not a cisco proprietary) depends on a cisco proprietary CEF?? If I use Non-Cisco routers in the mpls edge how come the labels get tagged for ip packets??
<<<<<<<<<Taken from Book>>>>>>>>>>>>>
MPLS Fundamentals - Luc De Ghein
Why Is CEF Needed in MPLS Networks?
Concerning MPLS, CEF is special for a certain reason; otherwise, this book would not explicitly
cover it. Labeled packets that enter the router are switched according to the label forwarding
information base (LFIB) on the router. IP packets that enter the router are switched according to
the CEF table on the router. Regardless of whether the packet is switched according to the LFIB
or the CEF table, the outgoing packet can be a labeled packet or an IP packet
<<<<<<<<>>>>>>>>>>

Answer 1

Hello Bava,
the key point is that LDP or RSVP TE are able to generate distribute labels for FECs but they do not create the FECs from stratch.
FEC = Forwarding Equivalent Class
a destination IP subnet is a typical FEC.
cisco MPLS code takes advantage of the work done by CEF and uses as input data the FIB (Forwarding Information Base) mantained by CEF, to build the LFIB that is the table where for each FEC there is an association with a label taken from the local node label space.
The work done by CEF is not so different from what is needed by MPLS: the biggest difference is that the CEF table is kept local and not exported to any other device. MPLS FEC/label bindings are advertised.
In MPLS frame mode the labels are distributed in unsolicited downstream mode.
Unsolicited means that the label/FEC association is buiilt based on the topology FIB instead  of waiting for some device to ask a label for the FEC.
downstream means the labels are sent in the opposite direction of that used by traffic.
Other attributes are:
indipendent : means each LSR is free to create its own FEC/label association before receiving the label from the edge LSR that owns the prefix or from a device that is nerarest to the IP subnet (upstream)
liberal retention: the device will keep note of labels advertised by neighbors even if they are currently not on the best path. This can be seen in the output of show mpls ldp binding and allows for faster recover in case of failure of the best path.
The unsolicited and liberal retention in standard frame mode comes from the relatively big label space (roughly one milllion labels)
This was not possible in MPLS cell mode where the label space was small. So MPLS cell mode used on demand downstream label binding and no retention.
Also MPLS allows for label stacking = use of multiple levels of MPLS Label for services like L3 VPNs
to be noted other implementations are different in some aspects and each vendor has its internal tecnique to build a table of FECs to be used as starting point for MPLS code.
For example indipendent label/FEC mapping has its own drawbacks it may be better to wait for a label to be received from a device upstream = nearest to the IP prefix in order to ensure the path is end to end.
Hope to help
Giuseppe