What They Never Tell You – The Secret of Ring
While most industrial Ethernet switch vendors compete for the speed of recovery time of their redundant ring, they never mention about what happens when you restore a broken ring.
Korenix’s innovation of Seamless Restoration brings a revolution to redundant ring technologies. The patented mechanism has 0 ms restoration time without causing any looping problem, topology change, and packet losses when recovering a failed ring to its original redundant-guarantee state.

The Ethernet Switching Principle
One important feature of Ethernet switch to make frame forwarding more efficiently is the ability of Address Learning. While frames are sent with source addresses and destination addresses, switches maintain a lookup table, named MAC address table, to track at which port a source address locates. Any unicast frame with a destination address that a switch has learnt in the lookup table, it can be forward to the correct port where the receiver locates without flooding to all ports. For multicast frames, switches use IGMP snooping to learn at which port there is device interested in the multicast traffic. However, if a frame’s destination address has not been learnt yet, called unknown unicast and unknown multicast, the frame is flooded to all ports. Flooding frames lead to broadcast storm which crashes the network if there is a loop in the topology. To eliminate loops in the network is a goal of some network protocols, such as standard STP/RSTP or proprietary redundant ring protocols.

How Ring Recovers from a Failure

1. RM has one of its ring paths blocked to provide redundant capability.	2. A link fails and the ring is broken into two parts. Packets are lost.
3. RM resumes the network connectivity by activating its redundant link. Topology changes.

What They Never Told You – The Risk of Ring Restoration

1. A broken ring.	2. A staff repairs the failed link. Topology changes and the network loops.
3. RM blocks one of its ring paths. The ring returns to the redundant-guarantee state.

Figure 2 Restoration procedure of a ring

When the ring is broken, it requires the staffs to fix the failed link. The procedure that RM regains its redundant capability is called Restoration and the duration is called Restoration Time.

	Restoration results in a network topology change.

Restoration brings the ring back to the redundant-guarantee state. However, it results in a topology change. A topology change alters the switching paths. Switches must clear their MAC address table and re-learn how to forward frames correctly. Network protocols, such as IGMP, GVRP, and GMRP, should also re-learn multicast, VLAN configuration after a topology change. During this period, may up to tens of seconds, all the unknown unicast and multicast traffic flood everywhere and degrade the network quality.
Both ring recovery and restoration result in topology change. The network suffers from the unstable status. This situation gets worse if restoration causes loop. Flooding unknown unicast and multicast traffic in a loop overwhelms the network in a blink.

0 ms, Error-Free Solution – Seamless Restoration
Korenix introduces its solution, Seamless Restoration, to these problems. As the name implies, this mechanism will not cause any influences on the network during ring restoration. There is no packet loss, loop or topology change. The restoration time is 0. This mechanism eliminates any unstable status and guarantees the applications keep running smoothly, as shown in Figure 3 Korenix Seamless Restoration reduces downtime to minimum, which is most suitable for mission-critical applications.

Figure 3 Korenix Seamless Restoration reduces downtime to minimum See Korenix Industrial Ethernet Switches feature Seamless Restoration Technology: JetNet , JetRock, JetPoE