7 HSR Support

HSR (IEC 62439-3 Clause 5), like PRP supports high-availability, zero down time networks. But instead of two independent networks, HSR uses a ring topology. When a node needs to use the network, the node sends duplicate Ethernet frames in opposite directions on the ring. The 1st frame received by the destination node is passed to the application layer while the 2nd frame is dropped.

In a ring topology, the forwarding delay of each node is a critical. Forwarding delay is the time required to receive and then transmit the frame back out onto the ring. As network size increases, the number of nodes increases, and thus the total latency increases. To reduce latency to acceptable levels, high-speed cut-through switching is required.

Similar to a PRP network, HSR provides high-availability service with zero recovery time and can remain operational during maintenance or equipment replacement cycles. If a non-HSR compliant device needs to connect to the HSR network, a RedBox can be used.

One of the disadvantages of an HSR network is that special hardware is required to support the HSR cut-through switching, adding to the cost of the network. Historically, FPGAs have been used in HSR networks, but FPGAs tend to be expensive. Off-the-shelf processor options like the AM64x that can support high-performance cut-through switching are helping to reduce HSR deployment costs.

HSR is referenced in the IEC 61850 standard for use in electrical substations.

The AM243x SDK includes support for 100Mbps HSR and the AM64x SDK includes support for 1Gbps HSR. See Table 9-1 and Table 9-2 for details.

Figure 7-5 illustrates a source node sending two copies of the frame simultaneously in both directions on the ring. If there is a failure point in one of the connecting network links or nodes, there is still a path for one of the frames to arrive at the destination node.

Figure 7-1 HSR Network Example