Extreme Networks ExtremeWare XOS 11.3, Manual

Page: 75 / 698

Understanding Hitless Failover Support—Modular Switches Only
ExtremeWare XOS 11.3 Concepts Guide
75
Table 11: Protocol support for hitless failover
Protocol Behavior Hitless
Spanning Tree
Protocol (STP) STP supports hitless failover including catastrophic failure of the MSM without
interruption. There should be no discernible network event external to the box.
The protocol runs in lock step on both MSMs and the backup MSM is a hot
spare that can take over at any time with no impact on the network. Yes
Ethernet Automatic
Protection Switching
(EAPS) Master Mode
If the master MSM state is Complete, the secondary port is blocking, and the
backup MSM is in the Init state. The backup also blocks the secondary port. If
the master fails over, the backup transitions from the Init state to the
Complete state. No change in blocking occurs and there should not be any
data loss.
If the master MSM state is Failed, the secondary port is forwarding. The
backup MSM enters the Init state. If the master fails over, the backup sends
its continuous "Query Link Status PDU" every 3 seconds. Within these 3
seconds, the backup receives the Link Down PDU. It then sends a message to
the lower layers to flush the FDB, and sets the secondary port to forwarding. If
the master fails over, there will be a slight hit to traffic due to FDB being
flushed.
No
EAPS Continued Transit Mode
If the master MSM state is LinksUp, both ports are forwarding. The backup
MSM is also in the LinksUp state with both ports set to forwarding. If the
master fails over, there is no change in backup's state. There should be no
data loss.
If the master MSM state is LinksDown, one or both ports are not forwarding
traffic. The backup MSM is also in the LinksDown state with the same ports
set to forwarding/blocking. If the master fails over, there is no change in the
backup’s state. There should be no data loss.
If the backup MSM state is Preforwarding, and the master MSM is in the
LinksUp state, the backup receives a Health-PDU within 1 second and learns
that the ring is complete. The backup sends a message to the lower layer to
flush the FDB and sets the port to the forwarding state. If the master fails
over, there will be slight hit to traffic due to FDB being flushed.
No
EAPS Continued EAPS Shared Ports, Controller Mode
If the master MSM state is Ready, the backup MSM is also in the Ready state.
If the master fails over, there is no state change for the backup. There should
be no data loss.
If the master MSM state is Blocking, the backup MSM is also in the Blocking
state. If no other link goes down, the backup blocks the same ports as the
master. If the master fails over, there is no change in the backup’s state. There
should be no data loss. However, if some segment's link is down, the backup
may not know about it until it becomes the master. The backup sends a
message to the lower layer to unblock the port and flush the FDB. In this case,
there could be a slight hit to traffic due to FDB being flushed.
In the Blocking state, there is a rare corner case situation where breaking two
links and healing one of them might cause the "Active Open" port to no longer
be the smallest port. The backup MSM does not know about this because of
the sequence of breaking and healing. In this case, the backup’s behavior is
unpredictable.
In the Blocking state and the Root Blocker state, the backup MSM does not
know about the Root Blocker state. The backup learns about the Root Blocker
state only after becoming the master MSM (within 1 second). At that point the
new master flushes the FDB and block segments. If the primary fails over,
there is a slight hit to traffic due to the FDB being flushed.
No