During our first discussion of Industrial Ethernet rings(article here), we covered some of the key drivers behind the technology including the need for simpler machine wiring, than what is provided with star topologies, and the importance of having high availability machine networks. This is a critical need regardless of which Industrial Ethernet network is used.
In the first part of the series, we discussed how this is accomplished on EtherNet/IP. In the second part of our series, we will discuss the technologies supported on PROFINet® for creating high availability machine networks.
PROFINet RT and PROFINet IRT are different technologies with PROFINeT which are optimized around the specific performance needs of the application. Additionally, these networks have unique needs for media redundancy that fit with the network’s performance requirements, so we will cover both Media Redundancy Protocol and Media Redundancy Planned Duplication protocol.
On PROFINet RT, the ring topology is supported through the Media Redundancy Protocol (MRP), a direct derivative of the Hirschmann HiPER-Ring topology developed in 1998. The ring is managed by the Media Redundancy Manager while other devices on the network behave as Media Redundancy Clients. The Media Redundancy Manager is responsible for issuing test frames to assess the integrity of the network and re-configuring the network in the event of a ring failure. During this time the Media Redundancy Clients forward test frames along the ring and respond to instructions regarding reconfiguration of the network. Control devices can either be connected into switches that support these capabilities or the switches can be designed into the devices themselves through various embedded interfaces.
During normal operation, the Media Redundancy Manager issues test frames, normally at 20ms intervals, out both ports as shown in Figure 1. In this initial state, the ports are all set to forward traffic. If the test frames are received back on both ports, the Media Redundancy Manager makes the determination that the ring is in a closed state.
At this point, the Media Redundancy Manager opens the ring by closing its second port and instructing the adjacent switch to close its second port to all other traffic except test frames and other network management frames. Additionally, the Media Redundancy Manager transmits instructions to each Media Redundancy Client on how to configure their forwarding database (FDB). These two steps have the effect of creating a linear topology for normal data traffic, which is critical to prevent data traffic from circulating in an infinite loop while maintaining a ring for network integrity.
If a break occurs on the network, this is detected by the Media Redundancy Manager through the failure of the test frames to traverse the ring and reach their destination. Accounting for retries and the standard 20ms interval, this typically takes about 60ms. Additionally, the Media Redundancy Clients will communicate failure messages back to the manager if a failure is detected on its forwarding port. Once a failure is detected, the Media Redundancy Manager will open its blocked port and send topology reconfiguration instructions out both ports to the clients. The manager will begin sending data traffic out both ports once the topology has been reconfigured and the manager and clients flush and reconfigure their FDBs. Generally, worst case scenario for time to reconfigure is 200ms but faster recovery times are more often realized.
For PROFINet IRT applications, in addition to Media Redundancy Protocol, the Media Redundancy Planned Duplication protocol can be supported supervisor. The network management features of this mirror Media Redundancy Protocol in terms of verifying topology and detecting breaks. The key difference is that rather than blocking data to the second port of the Media Redundancy Manager, this port is opened to allow data issued from senders as well as test frames issued from the Media Redundancy Manager to traverse the network in both directions.
This has the effect of creating a bumpless transfer in the event of a break in the ring. Since data packets traverse the network in both directions, they reach their endpoints regardless of the state of the network. This is critical for applications that require extremely high network availability.
To prevent the data packets from circulating in an infinite loop, the receiving devices need to support the capability to evaluate the incoming packets on each port and determine whether a duplicate packet has been received. It consumes the most recent packet and does not forward duplicates along the ring. This requires the use of a real-time Ethernet controller for PROFINet such as the Siemens ERTEC ASIC.
One final note on MRPD: It is critical to keep in mind that all devices on a MRPD network must support MRPD. It will not coexist gracefully with MRP-based devices.