Deployment Guide

MAC Move
MAC addresses can move from one port to another port that is also a tagged member of an OF-VLAN. The flow will be set up with the
new MAC address learnt and the VLAN and DMAC entry learnt on the orignal port will be flushed out when the MAC aging timer expires.
SDN link protection using Group Type Fast Fail-over (FF)
SDN link protection using group type - FF provides single-hop link protection between two SDN switches. A fast fail-over (FF) group is
designed to detect and respond quickly to port failures. The port failure will be detected by a port-down event and trigger a switch to
another bucket (buckets are a set of actions and associated parameters, defined for groups) with the port up. The Group type “fast fail-
over” (FF) has a list of buckets (1…N) as shown in the figure below. Each of them have a list of actions and a watch port as a special
parameter and only one can be selected as active at a time. This group type requires live port-monitoring to determine the bucket
selection. The port-down event can be a link layer down or physical layer down failure.
The watch port monitors the up or down status of a specific port. The buckets in the group are evaluated for selection in the order defined
by the group configuration. The first bucket which is associated with a live port is selected as the active port but the remaining are not
selected even though they are active. The bucket in use will not be changed unless the “liveness” of its watch port transitions down. The
port going down will trigger the FF group to select the next bucket in the list whose watch ports are “up”. Each bucket is associated with a
single port.
In this example, if the bucket 1 watch port goes down, bucket 2 will be selected. The transition time during a fast fail-over is dependent on
switch search time to find a watch port that is up, within the group. There is no guarantee on the switch time during FF, however this will
be quicker than consulting the controller to handle port-down events and modifying the flow. The switch software makes decision on the
new watch port and will be much quicker than consulting the controller.
The maximum number of buckets supported on a group is 16, to ensure system performance. A single port can be a member in more than
one group. This feature is closely controlled by the PFC controller during configuration. The switch is responsible for the fast fail-over.
When a watch-port fails, fast fail-over is triggered by the switch and traffic is moved to the next operational port in the group, the order of
selection is defined at configuration. The port status update message is sent to the controller. The group is modified by the controller and
the corresponding port is deleted from the group. This operation is not performed at the same time on both nodes. Similarly when the port
recovers and is operationally up, PFC sends a group modify message to add the port to a group. After fail-over, the group modify message
is not sent by PFC at the same time to both nodes participating in link protection. However the traffic is not affected and is carried by the
new watch port at each end. In order to protect the traffic there must be at least two ports in the group. This feature has been verified
using 2 buckets with NEC PFC Controller version 6.2 by Dell.
12
Introduction