MPLS Forwarding Plane: Connection-Oriented or Connectionless?

Wanted to pose this question to the resident experts. Is MPLS Forwarding Plane traffic considered Connection-Oriented or Connectionless? I've googled this and found very little pointing one way or another. This is probably an oversimplification, but I've always been under the impression that the MPLS control plane traffic that establishes LSPs was connection-oriented because it relied on TCP to exchange label information. Conversely, I see MPLS Data plane traffic as connectionless because AFAIK there isn't a mechanism to retransmit a lost packet or even determine if a packet made it to its destination. Is this view accurate?

Whaddya think?

Find more posts tagged with

Save $250 on 2025 certification boot camps from Infosec!

Book now with code EOY2025

Button

Comments

reaper81

It's not ut to MPLS to decide if packets need to be transmitted. First, as you probably know LSPs are unidirectional. MPLS can carry pretty much anything, Ethernet, ATM and so on. This frame then contains the IP header and there should be a TCP or UDP payload. There is really no difference running MPLS than running IP.

MPLS can use tunnels for redundancy with MPLS TE but it has nothing to do with the packets sent from the clients.

vinbuck

I was under the impression that once MPLS encapsulated a packet to transport it across an MPLS core that it didn't care what was in the packet until the MPLS header is decapsulated when it exits the MPLS portion of the network. If there is a problem in the MPLS core that causes a packet to drop while it's in transport as MPLS, there isn't a mechanism to retransmit it as far as I know. Now I understand that a TCP communication between two hosts that traverses an MPLS core will account for packet loss due to the nature of the protocol, but i'm talking about packet loss while a packet is encapsulated as MPLS.

What started me on this track is a discussion on whether or not MPLS packets lost in an EoMPLS tunnel would be retransmitted if they were dropped due to congestion. I'm inclined to believe that they aren't but I have yet to come across any documentation that is definitive on this point.