Good choice, there are tons of online material.
You can start small, explaining how BGP operates and how MP-BGP extensions allow it to be used for virtually any application and how it can allocate and distribute labels and how RFC3107 works, (btw same principles are used in BGP to distribute VPN labels).
And then you can tackle the more advanced topics, i.e. the use cases for RFC3107 and how it can help scaling MPLS backbones endlessly (to virtually any number of PEs).
You have the good old use cases in scaling IGP in multi-AS MPLS and Carrier Supporting Carrier deployments
And then reinvented RFC3107 for hierarchical MPLS within a single AS –helping to scale IGP when 1000’s of eNodeB (5G cells) are actually MPLS PEs (The buzzword for this use case is Unified/Seamless MPLS).
Also don’t forget to mention some of the drawbacks of this solutions –mainly convergence speed, since by default BGP is much slower than IGP, but there is BGP PIC Edge/Core to help speed things up.
But even BGP PIC core can’t help in some cases –as it depends on how fast the information about a failure of an egress PE can reach ingress PEs –and if ingress PEs are in different AS we have to rely on slow propagation via BGP (instead of fast IGP) for this purposes though there are other solutions promise to help with the restoration locally within the AS by a P router –until PEs in remote ASes are being notified.
https://tools.ietf.org/html/draft-minto-2547-egress-node-fast-protection-03
https://www.google.com/patents/US9525619
-there where many drafts related to this topic, but I’m not sure actually if any of these solutions can work with hierarchical MPLS deployments, i.e. RFC3107.
adam
netconsultings.com
::carrier-class solutions for the telecommunications industry::
adam
