The reason we want to do this is to save PE2 from doing one more lookup. It is a way of reducing CPU and memory usage on devices and making the process more efficient. Take a look at this post for more details:
As stated by Rene in the following post, the Forwarding Information Base (FIB), which is an integral part of the function of MPLS, is the CEF table. CEF is the fundamental switching path for MPLS. Without CEF, MPLS forwarding does not occur. For this reason, CEF is mandatory for MPLS to function.
The labels found within the LDP bindings table correspond to the LSR ID. The LSR ID is by default the highest IP address on a loopback interface. So in this case, yes, the labels will correspond to the Loopback interface IPs of the routers.
I assume you are referring to the Provider (P) routers and the Provider Edge (PE) routers used in MPLS, correct?
Keep in mind that the only difference between a P and PE router is the fact that the PE router has the added configuration of VRFs and customer connections. Otherwise the rest of the configuration is the same.
So really, it’s not so much the configuration of the router that makes it a P or PE, but the location it has in the topology. Imagine you have an MPLS network of just two routers connected to each other. They would both be PE routers since they connect customers, but they would also perform label distribution like a P router. Even so, they are called PE routers. Such a topology would have no P routers at all.
So really, there is no such thing as a combined P/PE router. Either you’re a PE router because you’re connected to customers, or you’re a P router because you’re not connected to customers. There’s no combining the two. Does that make sense?
BGP Labeled Unicast or BGP-LU is a technology that combines BGP with MPLS for efficient network routing.
In a typical IP network, the destination IP address is used to make routing decisions by each individual router along the path. In MPLS, however, it is the first router that makes a decision on the destination address and assigns a label to the packet. This label is then used by all subsequent routers to quickly and efficiently route the packet to its destination without having to independently determine the best route.
Now typically in an MPLS topology, these labels are exchanged using the Lable Distribution Protocol or LDP. More about that here:
BGP-LU can be viewed as an alternative to the use of LDP for the exchange of labels. BGP-LU uses BGP UPDATES to exchange these MPLS labels between routers, so it can also be used in a large-scale network where BGP is typically used, like ISP networks that compose sections of the internet. This allows for efficient, scalable routing over large distances.
Within the BGP UPDATE header, there is a field called Subsequent Address Family Identifier (SAFI). This identifier indicates what kind of information is carried by the update. A SAFI of 4 indicates that an MPLS label is included in the update using BGP-LU.
BGP-LU is often used in Service Provider networks for interconnecting different parts of the network or for providing MPLS-based services to customers. It also provides a way for network service providers to create VPNs and other services on top of their existing IP networks.
When would you use BGP-LU instead of LDP? Primarily when your MPLS topology spans multiple ASes. Typically, using BGP-LU will be able to exchange labels across AS borders much more easily than LDP would.
It’s important to remember that BGP-LU is not a separate protocol but a mechanism or feature that combines the capabilities of BGP and MPLS. The actual protocols used for communication remain BGP for routing information and MPLS for label information.
BGP-LU is defined in RFC 3107, and more about it in the context of Cisco can be found here:
This does sound strange indeed. However, this is how MPLS is intended to work. The LDP labels are used to denote specific paths within the MPLS core, so you will only see prefixes from within the core in the forwarding table. Indeed, the label that is added corresponds to the next hop IP rather than the actual destination network prefix.
A PE receives a packet from a CE, and if the destination is a remote site reachable via another PE router, the BGP table for that VRF will have an entry indicating the next hop (which is typically the other PE router). The corresponding label is added to the packet.
In the example you shared, the 184.108.40.206 network based on the BGP table, is reachable with a next hop of 220.127.116.11 (PE2). The label added to the packet that denotes the particlar path in the MPLS infrastructure is 17, which is the one added to the next hop address of 18.104.22.168.