Information About PRP
Parallel Redundancy Protocol (PRP) is defined in the International Standard IEC 62439-3. PRP is designed to provide hitless redundancy (zero recovery time after failures) in Ethernet networks.
Note |
PRP is supported on several Cisco Catalyst IE9300 Rugged Series Switches: IE-9320-26S2C-Eand IE-9320-26S2C-A beginning with Cisco IOS XE Cupertino 17,7,1, and IE-9320-22S2C4X-E, and IE-9320-22S2C4X-A beginning with Cisco IOX XE Dublin 17.12.1. |
To recover from network failures, redundancy can be provided by network elements connected in mesh or ring topologies using protocols like RSTP, REP, or MRP, where a network failure causes some reconfiguration in the network to allow traffic to flow again (typically by opening a blocked port). These schemes for redundancy can take between a few milliseconds to a few seconds for the network to recover and traffic to flow again.
PRP uses a different scheme, where the end nodes implement redundancy (instead of network elements) by connecting two network interfaces to two independent, disjointed, parallel networks (LAN-A and LAN-B). Each of these Dually Attached Nodes (DANs) then have redundant paths to all other DANs in the network.
The DAN sends two packets simultaneously through its two network interfaces to the destination node. A redundancy control trailer (RCT), which includes a sequence number, is added to each frame to help the destination node distinguish between duplicate packets. When the destination DAN receives the first packet successfully, it removes the RCT and consumes the packet. If the second packet arrives successfully, it is discarded. If a failure occurs in one of the paths, traffic continues to flow over the other path uninterrupted, and zero recovery time is required.
Non-redundant endpoints in the network that attach only to either LAN-A or LAN-B are known as Singly Attached Nodes (SANs).
A Redundancy Box (RedBox) is used when an end node that does not have two network ports and does not implement PRP needs to implement redundancy. Such an end node can connect to a RedBox, which provides connectivity to the two different networks on behalf of the device. Because a node behind a RedBox appears for other nodes like a DAN, it is called a Virtual DAN (VDAN). The RedBox itself is a DAN and acts as a proxy on behalf of its VDANs.
To manage redundancy and check the presence of other DANs, a DAN periodically sends Supervision frames and can evaluate the Supervision frames sent by other DANs.
Role of the Switch
IE-9320-26S2C-A, IE-9320-26S2C-E, IE-9320-22S2C4X-A, and IE-9320-22S2C4X-E switches implement RedBox functionality using Gigabit Ethernet port connections to each of the two LANs.
PRP or HSR on a IE9300 Stack
The Parallel Redundancy Protocol (PRP) offers significant advantages in facilitating redundancy with zero downtime. High-availability Seamless Redundancy (HSR) is similar to PRP but is designed to work in a ring topology. While the initial implementation was limited to standalone switches, recent advancements have enabled PRP or HSR to be utilized in stacked configurations.
Benefits of PRP or HSR on a IE9300 Stack
Deploying PRP or HSR in a stacked setup introduces a node level redundancy within the network.
-
This enhancement mitigates the risk of single points of failure.
-
Ensures the system remains operational even if a stack member or the active switch fails.
-
The functionality and behavior of PRP and HSR remain consistent in both standalone and stacked configurations.
-
Provides seamless integration and reliability across the network.
For more information on Switch Stacks, see Managing Switch Stacks.
Note |
There are no changes to the functionality or behavior of PRP and HSR when implemented in a IE9300 stack compared to their standalone configurations. |
The following illustration displays a IE9300 stack as a RedBox.
Note |
A maximum of four-member stack can be configured with PRP or HSR as RedBox. |
Guidelines and Limitations
The following guidelines and limitations apply:
-
You can create any one of the following configurations:
-
HSR: Maximum of two rings
or
-
PRP: Maximum of two channels
or
-
One HSR ring and one PRP channel
-
-
PRP and HSR are supported on IE-9320-26S2C and IE-9320-22S2C4X only.
-
PRP and HSR on IE9300 stack is supported only if both the active and standby switches are FPGA-based SKUs.
-
Both ports of a channel and a ring must be on the same slot, that is, the primary and secondary interfaces must be on the same switch member.
-
When PRP or HSR is configured on the active unit and the switch goes down, it remains unavailable until the switch is restored.
Active-Standby Synchronization Mechanism Post-Switchover
Synchronization of the PRP or HSR to a redundant standby, supportis both incremental and bulk synchronization updates.
The behavior of PRP channel or HSR ring when a switch goes down, is as follows:
-
If the channel or ring is configured on the standby switch, it will synchronize with the previous states of the channel or ring from the last active configuration.
-
If the channel or ring is configured on the active switch, it will transition to a down state due to the slot being inactive. Once the slot is reactivated, the volatile FPGA will be reprogrammed with the previously configured values.
PRP Channels
PRP channel or channel group is a logical interface that aggregates two Gigabit Ethernet interfaces (access, trunk, or routed) into a single link. In the channel group, the lower numbered Gigabit Ethernet member port is the primary port and connects to LAN-A. The higher numbered port is the secondary port and connects to LAN-B.
The PRP channel remains up as long as at least one of these member ports remains up and sends traffic. When both member ports are down, the channel is down. The total number of supported PRP channel groups is 2 per switch. The interfaces that you can use for each group on each switch series are fixed, as shown in the following table.
PRP Channel Number |
IE9300 Series |
---|---|
PRP Channel 1 |
Gi1/0/21 (LAN-A) and Gi1/0/22 (LAN-B) |
PRP Channel 2 |
Gi1/0/23 (LAN-A) and Gi1/0/24 (LAN-B) |
Mixed Traffic and Supervision Frames
Traffic egressing the RedBox PRP channel group can be mixed, that is, destined to either SANs (connected only on either LAN-A or LAN-B) or DANs. To avoid duplication of packets for SANs, the switch learns source MAC addresses from received supervision frames for DAN entries and source MAC addresses from non-PRP (regular traffic) frames for SAN entries and maintains these addresses in the node table. When forwarding packets out the PRP channel to SAN MAC addresses, the switch looks up the entry and determines which LAN to send to rather than duplicating the packet.
A RedBox with VDANs needs to send supervision frames on behalf of those VDANs. For traffic coming in on all other ports and going out PRP channel ports, the switch learns source MAC addresses, adds them to the VDAN table, and starts sending supervision frames for these addresses. Learned VDAN entries are subject to aging.
You can add static entries to the node and VDAN tables as described in x. You can also display the node and VDAN tables and clear entries. See y and z.
VLAN Tag in Supervision Frame
Cisco Catalyst IE9300 Rugged Series Switches support VLAN tagging for supervision frames. PRP VLAN tagging requires that PRP interfaces be configured in trunk mode. This feature allows you to specify a VLAN ID in the supervision frames for a PRP channel.
In the example configuration below, PRP channel 1 interface is configured in trunk mode with allowed VLANs 10 and 20. Supervision frames are tagged with VLAN ID 10. RedBox1 sends Supervision frames on behalf of VDANs with the PRP VLAN ID, but the regular traffic from VDANs goes over the PRP channel based on the PRP trunk VLAN configuration.
See Configuring PRP Channel with Supervision Frame VLAN Tagging for configuration information.