Brocade TurboIron 24X Series Configuration Guide - Manual (Page 322)

288 Brocade TurboIron 24X Series Configuration Guide
53-1003053-01
Metro Ring Protocol (MRP)
Alarm RHP
Previously, detection of MRP ring breaks was completely timer based. An absence of Ring Health
Packets (RHP) for a period of 3 "hello times" indicated to the MRP master that the ring is broken.
This initiated the transition to a topology change as described in the previous section. The
convergence time associated with such an event could take several hundreds of milliseconds.
Now, each MRP node is made a more active participant in detecting link failures. When a link is
detected to be down by its "downstream" neighbor, a special packet (called the Alarm RHP packet)
is sent to the MRP master, indicating that the link is down.
This MRP packet is sent from the MRP member to the MRP master only when the secondary link
goes down and it is sent on the primary link. The destination MAC address in the packet is the ring
MAC address. This allows the packet to be hardware forwarded all the way to the MRP master.
When the Master switch in the ring receives this packet, it is notified of a break in the ring. At that
point, the secondary interface is immediately transitioned from "Blocked" to "Forwarding" .
NOTE
The Alarm RHP packet is only sent by the secondary link owner ring to prevent multiple MRP masters
going into forward where shared rings are configured.
Operation of the MRP alarm RHP is as follows.
When the link between Switch B and Switch C fails, the "downstream" neighbor (Switch C) detects
the failure of the link and triggers corrective action. The following is the complete sequence of
events that occurs.
• The downstream neighbor (Switch C) detects a link down event of the link between Switch B
and Switch C.
• Switch C sends a single RHP packet with a special Alarm bit set. The RHP packet is sent in the
same direction of flow as that of the normal RHP packets (i.e. on the link to Switch D)
• Switch A receives the special RHP packet (on the secondary interface) that was sent by Switch
C. It is now aware that the ring is broken even though the dead_interval may not have expired.
• Switch A immediately transitions its secondary interface (previously in Blocked state) to the
Forwarding state.
• RHP packets continue to be sent on the primary interface by Switch A to detect if the ring has
been healed.
From a user perspective, there is no difference in the behavior of the ring. The only noticeable
difference is a rapid convergence in the event of ring failure. There is no CLI command required to
enable this feature.