VPLS PW Redundancy
Background
To protect against failures and improve reliability, a redundant provider edge (PE) is often deployed for a service. If a redundant PE is provided for a virtual private wire service (VPWS) or virtual private LAN service (VPLS), two pseudo wires (PWs) are deployed for PW protection. This mechanism is called PW redundancy.
PW redundancy, specified in draft-ietf-pwe3-redundancy-bit-04, can effectively increase devices' switchover efficiency to reduce service interruption and improve network reliability.
PW redundancy, which is widely used for point-to-point services on VPWS networks, can be used on VPLS networks because point-to-multipoint VPLS services can be considered as point-to-point services for each point.
PW redundancy used on VPLS networks is called VPLS PW redundancy. It helps rapidly converge VPLS networks and reduce service interruption.
Related Concepts
In Figure 1, VPLS PW redundancy protects service traffic transmission between customer edge 1 (CE1) and CE2 on the VPLS network.
Currently, VPLS PW redundancy can work in Master/Slave mode, specified on PE1. PE1 determines whether a local PW is in the primary or secondary state based on preset forwarding priority.
PEs on the two ends of a PW protection group must negotiate the PW states to ensure that they select the same PW to transmit packets.
Primary/Secondary: defines the PW forwarding priority. The PW forwarding priorities can be configured, and a smaller value indicates a higher priority. A PW with the highest priority is the primary PW.
PW forwarding priorities take effect only when PE1 uses PW redundancy in Master/Slave mode. In Master/Slave mode, PE1 instructs PE2 and PE5 to change the forwarding status of PWs to be the same as those of PWs on PE1. In Independent mode, the master and backup status of PE2 and PE5 determines forwarding priority of local PWs.
Active/Standby: defines the PW forwarding status and cannot be configured. Only active PWs are used for forwarding traffic. Standby PWs can only receive traffic.
Active/Inactive and Primary/Backup are terms used by Huawei that have the same meaning with Active/Standby defined in draft-ietf-pwe3-redundancy-bit-04. They all indicate the PW forwarding status.
Implementation
To ensure the same forwarding capability, two PEs must select the same PW to forward service data when the PW redundancy protection mechanism is used. In addition, only one PW in a PW protection group can be in the working status. To implement these functions, a signaling protocol is required.
RFC specifies the PW Status TLV to transmit the PW forwarding status. The PW Status TLV is transported to the remote PW peer using a Label Mapping or Notification message. The PW Status TLV is a 32-bit status code field. Each bit in the status code field can be set individually to indicate a PW forwarding status. PW redundancy introduces a new PW status code 0x00000020, which indicates "PW forwarding standby".
Forwarding priorities (Primary or Secondary) must be configured for PWs that back up each other. The highest priority PW will be selected as the primary PW to forward traffic. The remaining PWs will be in the Secondary state to protect the primary PW.
Currently, only one secondary PW can be configured for a primary PW.
The forwarding status of a PW determines whether the PW is used to forward traffic. The PW forwarding status depends on the following:
- Local and remote PW signaling status of a PW: A PE monitors the local signaling status and uses PW redundancy signaling to obtain remote signaling status from a remote PE.
- PW redundancy mode: Master/Slave or Independent mode is specified on PE1.
- PW forwarding priority: PW forwarding priorities (Primary or Secondary) are specified on PE1.
Figure 1 shows that VPLS PW redundancy is configured on PE1. In normal cases, all local and remote PW signaling states on PE1 are Up. PEs at the two ends of a PW in different VPLS PW redundancy modes use different methods to select the same PW for transmitting user packets.
In Master/Slave mode, PE1 determines local PW forwarding states based on preset forwarding priorities and inform PE2 and PE5 of the PW forwarding states.
PE2 and PE5 do not support to determine their PW forwarding states based on the received PW primary and secondary states.
In Independent mode, PE1 determines local PW forwarding states based on the forwarding states learned from PE2 and PE5; PE2 and PE5 determine their PW primary and secondary states based on signaling, which can be enhanced trunk (E-Trunk), enhanced automatic protection switching (E-APS), or Virtual Router Redundancy Protocol (VRRP), and notify PE1 of the forwarding states.
In both Master/Slave and Independent modes, if a primary PW is faulty, it becomes Inactive and its secondary PW becomes Active. PW-side faults do not affect the AC status. If AC-side faults occur (for example, a PE or an AC link is faulty), the PW primary and secondary states in Independent mode will change because the states are determined by the master and backup states of the dual-homing devices; the PW primary and secondary states in Master/Slave mode will not change because they are determined by the PW.
VPLS PW redundancy is similar to VPWS PW redundancy, with the exception that a virtual switch instance (VSI) has multiple PWs to different PEs. These PWs form various PW groups. PW switching in one group does not affect other PW groups.
Derivative Function
In addition to protection against network faults in real time, VPLS PW redundancy allows users to manually switch traffic between PWs in a group during network operation and maintenance. For example, if a device providing a primary PW needs to be maintained, a user can switch traffic to the secondary PW and switch it back to the primary PW after the maintenance.
The interval between a switchover and a revertive switchover must be at least 15s.
Usage Scenarios
VPLS PW redundancy can be used on hierarchical virtual private LAN service (HVPLS) networks as well as VPLS and virtual leased line (VLL) interconnected networks. These two types of networks can bear any services. However, service deployment suggestions in the two networking modes are as follows:
- HVPLS networks are suitable for bearing multicast services, such as Internet Protocol television (IPTV) services, because HVPLS networks can save VPLS core network bandwidth.
- VPLS and VLL interconnected networks are suitable for bearing unicast services, such as high-speed internet (HSI) and voice over IP (VoIP) services, because PEs on a VLL do not need to learn user MAC addresses.
VPLS PW redundancy can also be used to improve reliability of existing networks. On the VPLS network in Figure 1, CE1 communicates with CE2, CE3, and CE4 through PWs established between VSIs on PE1 and PE2, PE3, and PE4.
As services develop, services between CE1 and CE2, and between CE1 and CE3 require high reliability. Services between CE1 and CE4 do not require high reliability.
To meet the reliability requirements, PE5 and PE6 are deployed on the VPLS network to provide VPLS PW redundancy protection for PE2 and PE3. In addition, multiple PW groups to peer PEs are configured in one VSI on PE1. Links between CE1 and CE4 remain unchanged.
VPLS PW redundancy protects services against failures on the network, ACs, or PEs without affecting existing services, improving network reliability.
VPLS PW redundancy can be provided for the desired services without affecting services on other PWs, which reduces costs and maximizes profits.