Configuring LDP-based Virtual Private Wire Service

Before you start the hands-on part of this module, you should load the appropriate configuration and verify that the testbed is up and running by executing the corresponding robot file:

In order to get a better understanding, the lab setup is shown in the picture below.

In module Configuring EVPN Virtual Private Wire Service, we defined a Virtual Private Wire Service (VPWS) using EVPN-based pseudowires. In this module, we present an alternative approach that uses Label Distribution Protocol instead of BGP as the signaling protocol to establish and maintain pseudowires.

An LDP pseudowire enables service providers to emulate a point-to-point Layer 2 connection over an MPLS network. These pseudowires are built on top of the existing MPLS infrastructure and use extensions to LDP to signal and maintain virtual circuits between PE routers. For pseudowire setup, LDP relies on a targeted LDP session, where the TCP session is established between the two PEs using loopback addresses, i.e., the LDP neighbors do not have to be directly connected to each other.

Additionally, RFC 4447 introduces two new FEC elements: the pseudowire ID FEC element and the generalized pseudowire ID FEC element, which help the PE routers identify the endpoints of a pseudowire.

As a pseudowire is a point-to-point connection, it links two entities together. These endpoints are referred to as attachment circuits and represent the Ethernet interfaces connected to the CE device. The attachment circuit may either be an Ethernet port or a dedicated VLAN on an Ethernet port. Note that while the concept of pseudowires applies to other layer-2 technologies like ATM or Frame Relay, we will focus on Ethernet, as modern white-box switches typically only support Ethernet interfaces.

Once a pseudowire is configured and associated with an attachment circuit, the PE routers exchange LDP Label Mapping messages across the targeted LDP session that has been previously set up. An LDP Label Mapping message contains an FEC TLV, a Label TLV, and optional parameter TLVs. Typically, each FEC TLV consists of a set of FEC elements. However, for pseudowire setup and maintenance, the FEC TLV will contain exactly one FEC element.

In our example, the pseudowire is configured with a static pseudowire ID of 100 on both sides. Therefore, the PE routers use FEC 128, which includes the pseudowire ID. The LDP Label Mapping message also contains the Label TLV, which advertises the MPLS label that identifies the pseudowire and the label that the remote PE router needs to push onto the label stack.

The PW label bindings are distributed using LDP’s downstream unsolicited mode, meaning that the PE routers advertise the label to the remote PE router without waiting for a specific request from the downstream router.

When two PE routers have established a pseudowire, Ethernet traffic can be carried across the MPLS core network. An Ethernet frame—either untagged or tagged—is received on an incoming interface associated with the pseudowire. The local PE router pushes the pseudowire label, learned from the remote PE router, onto the frame and extracts the next-hop information, which is recursively looked up in the routing table, resulting in the pushing of a transport label. The frame is then forwarded to the next router along the LSP towards the downstream PE router.

Transit P routers are not aware of the VPWS and thus only act on the transport label, either swapping it or performing penultimate hop popping. The egress PE router receives a frame containing only the PW label. It will perform an MPLS lookup, which results in forwarding the packet to the attachment circuit.

In order to configure LDP-based pseudowires in RBFS, you must first set up an MPLS core network. You can choose any IGP or MPLS signaling protocol. Just because we are using LDP for pseudowire setup doesn’t mean you need to use LDP to establish LSPs as well—although there is some synergy. Before diving into the details, let’s go over a quick checklist.

Now that we have a targeted LDP session up and running between the PE routers, we need to configure the attachment circuit. Since the attachment circuit is a Layer 2 interface that sends and receives frames without inspecting their content, there’s no need to configure an IP address. Instead, the interface-type must be set to l2vpn-vpws.

In this example, we have configured two single-tagged attachment circuits. RBFS also supports untagged and double-tagged attachment circuits.

The pseudowire definition is configured under the set instance <instance> protocol ldp vpls-vpws command hierarchy, associating the local attachment circuit with the unique pseudowire ID and the remote PE router.

We can verify that the pseudowires are established. The prefix is display in BGP EVPN style. The most important part is the PW ID.

Further we can confirm that the LDP pseudowire is installed into the RBFS routing table.

The control word, defined in RFC 4385, is an optional 32-bit field inserted between the MPLS label stack and the Layer 2 payload. Its primary purpose is to prevent misinterpretation of the payload as an IP header by MPLS core routers, and it also helps detect packet misordering. The usage of the control word must be negotiated and agreed upon during signaling. The control word is an optional step in the pseudowire configuration. By default, the use of control word is disabled.

This module outlined the building blocks of LDP-based pseudowires. You should be able to configure an LDP-based VPWS.

If you have completed the exercise, you can check the results by executing