ITU G.8032 ERPS (Ethernet Ring Protection Switching)
Last updated
Last updated
G.8032 is an ITU-T standards-based Ethernet ring loop protection protocol. It is also known as ERPS (Ethernet Ring Protection Switching). G.8032 is an alternative to REP, and shares many similarities with Cisco’s propreitary ring protection protocol. I believe G.8032 converges even faster than REP because it uses CFM for link failure detection.
One big difference in G.8032 compared to REP, is that in G.8032 you pick the link that you want to block. This link is called the RPL (ring protection link). The device and port the link connects to is called the RPL owner and the RPL port, respectively. There is no such thing as an “edge port” in G.8032.
If a link in the ring fails, the devices on either end of the failed link send R-APS (Ring Automatic Protection Switching) messages indicating that the link has failed and to unblock the RPL port. This message is called an R-APS SF (Signal Failure) message.
If the link comes back up, the devices send R-APS NR (No Request) messages that indicate that the link is back up, and to block the RPL port again. The R-APS messages are basically on/off switches that turn the RPL link on and off depending on whether there is link failure in the topology. Just like REP, you can only have at most one single link fail at any given time.
Instead of PDUs, G.8032 relies on CFM to detect link failures. Therefore to run G.8032 you need to enable CFM on all nodes and configure a CFM domain. G.8032 uses CFM continuity check messages for which you can configure the interval as low as 3.3msec. This allows for extremely fast failover.
Just like REP, we cannot run G.8032 on virtual equipment. However we can examine the basic configuration steps to understand how it works.
Enable CFM and configure g8032 as a registered client of CFM. This is very similar to the way routing protocols register with BFD. When CFM detects the link down, it notifies g8032. efd stands for Ethernet Fault Detection.
Each device should have a unique mpid, and you should be able to use the same mpid on each of the two interfaces. (R1 can use mpid 1 on both interfaces, R2 can use mpid 2 on both interfaces, etc). By defining every interface as a MEP, it means that every CFM domain only spans a single link. Put another way, every link in the ring has its own independent CFM domain.
Create a g8032 ring profile.
Define the ring. You must associate up to two ports with the ring on every device. This is a basic rule of closed rings - every device has exactly two ports belonging to the ring. If the ring is an open ring, two devices will only have one port in the ring. I believe every device in the open ring will need to define the ring as an open-ring in this case.
