Ethernet Ring Protection
Link failure is often an unavoidable part of networking. However, there are methods of improving the reliability of a router or bridge network even when link failures occur. For example, SONET/SDH seal-healing rings are frequently used to add a level of robustness to router networks. This ring protection switching is now extended to Ethernet links. You can configure Ethernet ring protection for a series of two or more systems so that if one link fails, traffic is rerouted around the failure on the ring.
The basic idea of Ethernet ring protection is to use one specific link to protect the whole ring. This special link is the ring protection link (RPL). When all links are up and running, the RPL blocks traffic and remains idle. The RPL itself is controlled by the designated RPL owner node. There is only one RPL owner node on the ring and the RPL owner node is responsible for blocking the RPL interface under normal operating conditions. However, if a link failure occurs on the ring, the RPL owner node is responsible for unblocking the RPL interface and protection–switching the traffic on the alternate path around the ring. An Ethernet ring automatic protection switching (R-APS) messaging protocol coordinates the protection activities of all nodes on the ring. The APS blocks traffic over the failed link and unblocks traffic over the RPL.
When the failed link is repaired, the traffic reverts to its normal pattern. That is, the RPL owner blocks the RPL link and unblocks traffic over the cleared link.
Two or more nodes form a ring. Links between the nodes form a chain, with the last node also connecting the first. Every ring node therefore has two ports related to the ring, one in each direction. In this chapter, these directions are referred to as east and west.
Every node on the ring is one of two types:
RPL owner node—This node owns the RPL and blocks or unblocks the RPL as conditions require. This node initiates the R-APS message.
Normal node—All other nodes on the ring (that is, those that are not the RPL owner node) operate as normal nodes and have no special role on the ring.
In addition to roles, each node on the Ethernet ring can be in one of several states:
Init—The node is not yet participating in the ring.
Idle—The node is performing normally (there is no link failure on the ring). In this state, traffic is unblocked on both ring ports, except for the RPL owner node, which blocks the RPL port (the other RPL owner port is unblocked).
Protection—When a failure occurs on the ring, a normal node will have traffic blocked on the ring port that connects to the failed link. The RPL owner, if it is not at one end of the failed link, will then unblock the RPL port so both ports are active.
The R-APS protocol does not detect the number of RPL owner nodes configured on the ring. You must configure only one RPL and RPL owner per ring or protection switching will not work properly.
Ethernet ring protection only works when one link on the ring fails. Multiple link failures will break the ring and cause protection switching to fail.
Several restrictions apply to Ethernet ring protection:
The Ethernet ring protection configured as a single instance only works at the physical level (adjacent nodes must be directly connected). The ring protection operates at the interface (port) level and not at the VLAN level.
Manual (command-based) switching to protection mode is not supported.
Nonrevertive switching is not supported. When the link failure is cleared, traffic always returns to normal operation.
The interconnection of multiple rings for protection purposes is not supported.
You can configure Ethernet ring protection to optomize traffic load-balancing by using multiple ring instances. For more information about multiple ring instances, see Ethernet Ring Protection Using Ring Instances for Load Balancing