Introduction to the Virtual Leased Lines "VLL"

Definition
  • MPLS L2VPN
    The Multiprotocol Label Switching Layer 2 Virtual Private Network (MPLS L2VPN) transmits Layer 2 VPN services over an MPLS network. MPLS L2VPN enable operators to provide L2VPN services over different media, such as Asynchronous Transfer Mode (ATM), Frame Relay (FR), virtual local area network (VLAN), Ethernet, and Point-to-Point Protocol (PPP) in a unified MPLS network.
    Simply, the MPLS L2VPN indicates that Layer 2 data is transmitted transparently over an MPLS network. For the users, the MPLS network functions as a Layer 2 switched network through which Layer 2 connections can be set up between nodes. Layer 2 connections can be set up in virtual leased line (VLL) mode and virtual private LAN service (VPLS) mode.
  • VLL
    The VLL is an emulation of the traditional leased line service. It emulates the leased line over an IP network, and provides the asymmetrical digital data network (DDN) service at low costs. For users at both ends of a VLL, the VLL is similar to the traditional leased line. The VLL is a point-to-point virtual private wire technology that can support almost all the link layer protocols. The VLL can be implemented in the following modes:
    • Circuit Cross Connect (CCC): It is a mode of implementing the L2VPN through static configuration.
    • Static Virtual Circuit (SVC): It is a mode of implementing the MPLS L2VPN. The SVC is similar to the Label Distribution Protocol (LDP) L2VPN. The difference is that LDP is not used as the signaling protocol for transmitting VC labels or link information, whereas VC labels are manually configured on the SVC.
    • Martini: It implements the MPLS L2VPN by using LDP as the signaling protocol for transmitting the VC information.
    • Kompella: It implements the MPLS L2VPN by using the Border Gateway Protocol (BGP) as the signaling protocol for transmitting the VC information.
    • Pseudo-Wire Emulation Edge to Edge (PWE3): It is an extension of Martini mode and a technology for end-to-end Layer 2 service transmission.
  • VPLS
    VPLS uses the PSN to connect multiple Ethernet LAN segments and thus these segments can work as one LAN. VPLS is also called transparent LAN service or virtual private switched network service (VPSNS).
    Different from the point-to-point service of the common L2VPN, VPLS enables the service provider to offer Ethernet-based multipoint service to users through an MPLS backbone network.

Purpose

  • Extended network functions and service capabilities of operators
    Operators can provide MPLS L2VPN services over only one network. In addition, operators can use enhanced technologies related to MPLS, such as traffic engineering (TE) and Quality of Service (QoS), to provide users with different classes of services to meet users' requirements.
  • Higher scalability
    In an ATM or FR network that MPLS is not enabled, VCs provide the L2VPN service. For each VC, the provider edge (PE) devices and provider (P) devices in the network need to maintain the complete VC information. Then, when PEs of the operators are connected to multiple costumer edge (CE) devices, multiple VCs are created. Therefore, PEs and P devices must maintain information about multiple VCs. The MPLS L2VPN, however, can adopt label stacking to multiplex multiple VCs in a label switched path (LSP). Therefore, P devices only need to maintain information about one LSP. This improves scalability of a system.
  • Separation of administrative responsibilities
    In the MPLS L2VPN, operators provide only Layer 2 connectivity while users are responsible for Layer 3 connectivity such as routing. Therefore, route flapping caused by incorrect configurations does not affect stability of operators' networks.
  • Ease of configuration
    In a traditional L2VPN, if N CEs are connected through the full mesh topology, you must set up N - 1 permanent virtual circuits (PVCs) on each CE to connect other CEs. Then, the number of PVCs in the entire network is N x (N - 1)/2. Especially when a CE is added, you need to additionally set up N PVCs on the new CE, and then create a PVC connecting the new CE on each of original N CEs. In an MPLS L2VPN in Kompella mode, however, the number of CEs configured initially is excessive. When adding a CE, you need to modify the configuration of only the PE that is directly connected to the new CE rather than other PEs.
  • Support for multiple protocols
    Operators provide only Layer 2 connections; therefore, users can use any Layer 3 protocol such as IPv4 and IPv6.
  • Smooth network upgrade
    The MPLS L2VPN is transparent to users; therefore, when operators upgrade networks from traditional L2VPNs such as ATM and FR networks to MPLS L2VPNs, users do not need to perform any configuration. The network upgrade does not affect user services except for data loss in a short period during the switchover.

VLL in CCC Mode

In CCC mode, the L2VPN is implemented through the static configuration.
The CCC mode is applicable to an MPLS network of small size and simple topology, which requires the manual configuration of the administrator. No control packet needs to be transmitted and no signaling negotiation is performed; therefore, the MPLS L2VPN in CCC mode consumes a few resources and is easy to configure.

Classification of CCCs

CCC connections can be classified into the following types:
  • Local connection: indicates the connection of two local CEs. That is, the two CEs are connected to the same PE. Similar to a Layer 2 switch, the PE can directly transmit packets without a static LSP to be set up.
  • Remote connection: indicates the connection of the local CE and the remote CE. The two CEs are connected to PEs. In this case, static LSPs need to be set up to transmit packets from one PE to another PE.

Topology in CCC Mode

The MPLS L2VPN in CCC mode supports both the remote connection and the local connection. Figure 1 shows the topology of the MPLS L2VPN in CCC mode.
Figure 1 MPLS L2VPN topology in CCC mode 

As shown in Figure 1, Site 1 and Site 2 of VPN1 are interconnected through the remote CCC connection (the blue dashed line). Two static LSPs need to be set up between Site 1 and Site 2. One LSP is set up from PE1 to PE2, indicating the LSP from Site 1 to Site 2; the other LSP is set up from PE2 to PE1, indicating the LSP from Site 2 to Site 1. The two blue dashed lines constitute a bidirectional VC, that is, a CCC remote connection that is similar to the traditional L2VPN connection provided for customers.
As shown in Figure 1, Site 1 and Site 2 of VPN2 are interconnected through the CCC local connection (the black dashed line). PE3 that Site 1 and Site 2 access functions as a Layer 2 switch. LSP tunnels need not to be set up between CEs. Data of different link types, such as VLAN, Ethernet, FR, ATM AAL5, PPP, and HDLC, can be directly exchanged.
The advantage of the CCC mode is that no label signaling is needed to transmit L2VPN information so long as the Internet service provider (ISP) network supports MPLS forwarding. In addition, QoS can be guaranteed because the LSP in MPLS L2VPN in CCC mode is private.

VLL in Martini Mode

Definition

In Martini mode, you can set up a point-to-point link to implement the L2VPN and adopt LDP as the signaling protocol to transmit VC information. In this manner, the MPLS L2VPN is implemented.
The Martini mode uses double labels. The inner label is exchanged through extended LDP; the outer label is the tunnel label.
In Martini mode, multiple VCs can be set up over one LSP between two PEs. In addition, PEs store only a small amount of L2VPN information such as mappings between VC labels and LSPs. P devices, however, do not store any L2VPN information. Therefore, Martini features excellent scalability. To add a VC, you only need to configure a unidirectional VC on each PE of both ends. This does not affect the running of a network.
Compared with Kompella, Martini that adopts LDP rather than BGP as the signaling protocol is independent of the periodic refresh mechanism; therefore, faults can be detected faster in the Martini VLL.

Basic Concepts

In Martini mode, the VC type and the VC ID are used together to identify a VC between CEs.
  • VC type: indicates the encapsulation type of a VC, for example, VLAN, ATM, and PPP.
  • VC ID: identifies a VC. All VCs of the same type on a PE must have a unique ID.
The PEs that are connected to two CEs exchange VC labels through LDP, and the CE is bound to the peer CE according to the VC ID. In addition, the VC that transmits Layer 2 data can be successfully set up if the following conditions are all met:
  • The physical status of AC interfaces is Up.
  • The tunnel between PEs is successfully set up.
  • Labels are exchanged between PEs and each label is bound to a VC ID.
In Martini mode, the outer VC label is used to transmit data of each VC in an ISP network. The inner VC label is used to identify user data. Therefore, an LSP in the ISP network can be shared by multiple VCs.
The outer tunnel is used to transmit VC data across the ISP network; therefore, the outer tunnel can be an IP-encapsulated tunnel, for example, a GRE tunnel.
In deployment of the VLL in Martini mode, LSPs must be automatically set up in the ISP network. Therefore, the ISP network must support MPLS and MPLS LDP. If MPLS LDP is not supported, the GRE tunnel can be used instead.
The VLL in Martini mode supports graceful restart (GR). In this manner, after the device performs the switchover, the VC labels remain unchanged. During the switchover, the VC keeps Up. The packet forwarding on the VC is not affected by the switchover.
check the configuration for Martini VLL here

Topology in Martini Mode

The MPLS L2VPN in Martini mode only supports the remote connection rather than the local connection. Figure 2 shows the topology of the MPLS L2VPN in Martini mode.
Figure 2 MPLS L2VPN topology in Martini mode 

As shown in Figure 2, Site 1 and Site 2 of VPN1 are interconnected through the remote Martini connection (the black dashed line). Site 1 and Site 2 of VPN2 are also interconnected through the remote Martini connection (the blue dashed line). VPN1 and VPN2 are interconnected through two LSPs in an ISP network. They can also multiplex an LSP for interconnection.

VLL in SVC Mode


Definition

In Martini mode, LDP is used for exchanging VC labels. In SVC mode, LDP is not used, and the inner label is assigned manually on the PE according to the VC ID. Therefore, the SVC mode can be regarded as the simplified Martini mode.
In the SVC VLL, VC labels are manually configured without VC label mapping. Therefore, LDP is not needed for transmitting VC labels.

Topology in SVC Mode

The setup method of the outer label (specifying a public tunnel) in SVC mode is the same as that in Martini mode. The inner label is specified manually during the VC configuration. PEs do not need the signaling protocol to transmit VC labels. Therefore, the network topology and packet exchange process in SVC mode are the same as those in Martini mode.
When creating a static Layer 2 VC connection in SVC mode, you can specify a tunnel type such as the LDP LSP, constraint-based routing label switched path (CR-LSP), and GRE tunnel through the tunnel policy, and set up tunnels to support load balancing. The SVC mode supports the inter-AS L2VPN in multi-hop mode, but does not support the local connection.

VLL in Kompella Mode


Definition

The VLL in Kompella mode adopts BGP as the signaling protocol to transmit Layer 2 information and VC labels between PEs.
The Kompella VLL uses VPN targets to control the receiving and sending of VPN routes, which improves flexibility of the VPN networking. In Kompella mode, VC labels are assigned through a label block that is pre-allocated for each CE. The size of the label block determines the number of connections that can be set up between the local CE and other CEs. In addition, in Kompella mode, additional labels can be assigned to L2VPNs in the label block for expansion in the future. PEs calculate inner labels according to these label blocks and use the inner labels to transmit packets. The VLL in Kompella mode has good scalability and supports both the remote connection and the local connection.

Topology in Kompella Mode

The VLL in Kompella mode supports both the remote connection and the local connection. Figure 3 shows the VLL topology in Kompella mode.
Figure 3 VLL topology in Kompella mode 

As shown in Figure 3, Site 1 and Site 2 of VPN1 are interconnected through the remote Kompella connection (the black dashed line). Site 1 and Site 2 of VPN2 are interconnected through the local Kompella connection (the blue dashed line).
BGP auto-discovery enables the Kompella mode to better support complicated topologies.

7 comments: