Mrp Failover Performance - GE PACSystems RX3i User Manual

6.1.1

MRP Failover Performance

A network using Media Redundancy Protocol recovers from a ring failure within 80 milliseconds when running
at 100 Mbps full duplex with default values. Actual failover time depends on the device responsiveness to
network disconnection and reconnection, number of devices in the ring, media speed, length of media, and
frequency of sending test frames over the network. Network recovery time is shorter with fewer devices, faster
media speed, and shorter media lengths. If all MRC devices provide LinkUp/LinkDown indications, the network
recovery time will be significantly shorter than 80ms. Third-party devices in the MRP ring may introduce
additional network recovery time.
When MRCs do not provide LinkUp/LinkDown detection, network recovery time also depends upon the test
packet timeout interval, which is the product of the Default Test Interval and the Test Monitoring Count plus 1
that have been set for the Media Redundancy Manager. These parameters determine the frequency for
network integrity checks and the number of failed integrity checks to allow before declaring a ring failure. For
an PACSystems PROFINET Controller acting as Media Redundancy Manager, these parameters are set as part
of the Proficy Machine Edition hardware configuration. (For example, the default test packet timeout interval is
20 ms × (3+1) = 80 ms.)
Each device in the ring must be able to detect the failure or recovery of a connection. When using third-party
devices in a redundant media ring, failover time will be affected by the performance of these devices. For
devices that do not detect the failure or recovery of a connection via LinkUp/LinkDown indication, the shortest
guaranteed failover time will be the test packet timeout interval, as described above.
For bumpless network recovery (without disturbing I/O communications to an I/O Device), the Update Rate for
the I/O Device should be configured to be greater than 1/2 of the network recovery time. This permits the ring
to be disconnected or reconnected without timing out the communication connection between the I/O Device
and its I/O Controller.
1000Base-T SFPs on a PNC001 module can introduce significant network recovery time. For best MRP
performance with a 10/100/1000 Mbps copper Ethernet MPR ring network, configure the RJ45 Ethernet ports
(Ports 1 and 2) as MRP ring ports instead of those SFP ports (Ports 3 and 4) with 1000Base-T SFPs.
Note: The 1000BaseT SFPs qualified for use with the RX3i PNC001 module can only detect a ring break when
running at 1 Gbps at the IEEE 802.3 Clause 40 standard requirement of 750ms. In order to experience
bumpless I/O at a 16ms I/O Update rate with these SFPs at 1 Gbps, the MRM must be configured with
a Test Packet Interval of 10ms and a Test Packet Count of 2.
GFK-2571N May 2018
Chapter 6. Redundant Media
129