SR-MPLS BE and LDP Communication

As an emerging technology, SR-MPLS has received more and more attention. SR-MPLS is introduced to simplify network deployment and management and reduce capital expenditure (CAPEX) and operating expenditure (OPEX). MPLS LDP is a mainstream tunneling technique that is widely used on bearer networks. Before LDP is completely replaced by SR-MPLS BE, LDP and SR-MPLS BE will coexist for a long time. Interworking between an LDP network and an SR-MPLS BE network, therefore, becomes a problem that must be addressed.

The SR-MPLS BE and LDP interworking technique allows both segment routing and LDP to work within the same network. This technique connects an SR-MPLS BE network to an LDP network to implement MPLS forwarding.

To implement the interworking between the LDP and SR-MPLS BE networks, the SR-MPLS BE network must have devices that replace SR-incapable LDP devices to advertise SIDs. Such devices are mapping servers.

Mapping server: supports mapping between prefixes and SIDs and advertises the mapping to a mapping client.
Mapping client: receives mapping between prefixes and SIDs sent by the mapping server and creates mapping entries.

Since LSPs are unidirectional, SR-MPLS BE and LDP interworking involves two directions: SR to LDP and LDP to SR.

SR-MPLS to LDP

Figure 1 describes the process of creating an E2E SR-MPLS-to-LDP LSP.

Figure 1 Process of creating an E2E SR-MPLS-to-LDP LSP
Click to enlarge

The process of creating an E2E SR-MPLS-to-LDP LSP is as follows: On PE2, an IP address prefix is configured. LDP assigns a label to the prefix. PE2 sends a Label Mapping message upstream to P3.
Upon receipt of the message, P3 assigns a label to the prefix and sends a Label Mapping message upstream to P2.
Upon receipt of the message, P2 creates an LDP LSP to PE2.
On P2, the mapping server function is enabled so that P2 maps an LDP label carried in the IP address prefix to a SID.
P2 advertises a Mapping TLV upstream to P1.
P1 advertises a Mapping TLV upstream to PE1.
PE1 parses the Mapping TLV and creates an SR LSP to P2.
P2 creates mapping between the SR and LDP LSPs.

During data forwarding, P2 has no SR-MPLS label destined for PE2 and encapsulates an SR-MPLS label to an LDP label based on the mapping between the prefix and SID.

LDP to SR-MPLS

Figure 2 describes the process of creating an E2E LDP-to-SR-MPLS LSP.

Figure 2 Process of creating an E2E LDP-to-SR-MPLS LSP

The process of creating an E2E LDP-to-SR-MPLS LSP is as follows: An IP address prefix is assigned to PE1 and a SID is set for the prefix. PE1 advertises the prefix and SID to P1 using an IGP.
Upon receipt of the information, P1 advertises the prefix and SID to P2 using an IGP.
Upon receipt of the prefix and SID, P2 creates an SR LSP to PE1.
On P2, proxy LDP egress is configured P2 maps a SID carried in the IP address prefix to an LDP label. Once proxy LDP egress is configured and the route to the peer is reachable, a local node sends a Label Mapping message upstream.
P2 sends a Label Mapping message upstream to P3, and P3 sends a Label Mapping message upstream to PE2.
PE2 parses the received Label Mapping message and creates an LDP LSP to P2.
P2 creates mapping between the SR and LDP LSPs.

During data forwarding, P2 has no LDP label destined for PE1 and encapsulates an LDP label to an SR-MPLS label based on the mapping between the prefix and SID.