MPLS Forwarding

MPLS Forwarding Process

Basic Concepts

Label operations involved in MPLS packet forwarding include push, swap, and pop:

• Push: When an IP packet enters an MPLS domain, the ingress node adds a new label to the packet between the Layer 2 header and the IP header. Alternatively, an LSR adds a new label to the top of the label stack.

• Swap: When a packet is transferred within the MPLS domain, a local node swaps the label at the top of the label stack in the MPLS packet for the label allocated by the next hop according to the label forwarding table.

• Pop: When a packet leaves the MPLS domain, the label is popped out of (removed from) the MPLS packet.

  A label is invalid at the last hop of an MPLS domain. The penultimate hop popping (PHP) feature applies. On the penultimate node, the label is popped out of the packet to reduce the size of the packet that is forwarded to the last hop. Then, the last hop directly forwards the IP packet or forwards the packet by using the second label.

  By default, PHP is configured on the egress node. The egress node supporting PHP allocates the label with the value of 3 to the penultimate hop.

Basic Forwarding Process

LSPs that support PHP are used in the following example to describe how MPLS packets are forwarded.

Figure 1-6 Basic MPLS forwarding process

As shown in Figure 1-6, the LSRs have distributed MPLS labels and set up an LSP with the destination address of 4.4.4.2/32. MPLS packets are forwarded as follows:

1. The ingress node receives an IP packet destined for 4.4.4.2. Then, the ingress node adds Label Z to the packet and forwards it.

2. When the downstream transit node receives the labeled packet, the node replaces Label Z by Label Y.

3. When the transit node at the penultimate hop receives the packet with Label Y, the node pops out Label Y because the label value is 3. The transit node then forwards the packet to the egress node as an IP packet.

4. The egress node receives the IP packet and forwards it to 4.4.4.2/32.

MPLS TTL Processing

This section describes how MPLS processes the TTL and responds to TTL timeout.

MPLS TTL Processing Modes

The TTL field in an MPLS label is 8 bits long. The TTL field is the same as that in an IP packet header. MPLS processes the TTL to prevent loops and implement traceroute.

MPLS TTL processing modes includes: Uniform and Pipe modes. By default, MPLS processes the TTL in Uniform mode. The two modes work as follows:

• Uniform mode

  When IP packets enter an MPLS network, the ingress node decreases the IP TTL by one and copies this new value to the MPLS TTL field. The TTL field in MPLS packets is processed in standard mode. The egress node decreases the MPLS TTL by one and maps this new value to the IP TTL field. Figure 1-7 shows how the TTL field is processed on the transmission path.

  Figure 1-7 TTL processing in Uniform mode
  

• Pipe mode

  As shown in Figure 1-8, the ingress node decreases the IP TTL by one and the MPLS TTL remains constant. The TTL field in MPLS packets is processed in standard mode. The egress node decreases the IP TTL by one. In Pipe mode, the IP TTL only decreases by one on the ingress node and one on the egress node when packets travel across an MPLS network.

  Figure 1-8 TTL processing in Pipe mode
  

In MPLS VPN applications, the MPLS backbone network needs to be shielded to ensure network security. The Pipe mode is recommended for private network packets.

ICMP Response Packet

On an MPLS network, when an LSR receives an MPLS packets with the TTL value of 1, the LSR generates an ICMP response packet.

In most cases, the received MPLS packet contains only one label and the LSR responds to the sender with the ICMP response packet using the first method. If the MPLS packet contains multiple labels, the LSR uses the second method.

The MPLS VPN packets may contain only one label when they arrive at an autonomous system boundary router (ASBR) on the MPLS VPN. These devices have no IP routes to the sender, so they use the second method to reply to the ICMP response packets.