Configuring Controller Redundancy And Clustering - Brocade Communications Systems RFS6000 System Reference Manual

Configuring controller redundancy and clustering

Configuration and network monitoring are two tasks a network administrator faces as a network
grows in terms of the number of managed nodes (controllers, routers, wireless devices etc.). Such
scalability requirements lead network administrators to look for managing and monitoring each
node from a single centralized management entity. The controller not only provides a centralized
management solution, it provides centralized management from any single controller in the
network without restricting or dedicating one controller as a centralized management node. This
eliminates dedicating a management entity to manage all redundancy members and eliminates
the possibility of a single point of failure.
A redundancy group (cluster) is a set of controllers (nodes) uniquely identified by group/cluster ID.
Within the redundancy group, members discover and establish connections to other group
members. The redundancy group has full mesh connectivity using TCP as the transport layer
connection.
Up to 12 controllers can be configured as members of a redundancy group to significantly reduce
the chance of a disruption in service to WLANs and associated Clients in the event of failure of a
controller or intermediate network failure. All members can be configured using a common file
(cluster-config) using DHCP options. This functionality provides an alternative method for
configuring members collectively from a centralized location, instead of configuring specific
redundancy parameters on individual controllers.
Configure each controller in the cluster by logging in to one participating controller. The
administrator does not need to login to each redundancy group member, as one predicating
controller can configure each member in real-time without "pushing" configurations between
controllers. A new CLI context called cluster-cli is available to set the configuration for all members
of the cluster. All controller CLI commands are considered cluster configurable.
In the following example, there are four controllers (WS1, WS2, WS3 and WS4) forming a
redundancy group. Each controller has established a TCP connection with the others in the group.
There is an additional CLI context called cluster-context. A user/administrator can get into this
context by executing a cluster-cli enable under the CLI interface (future releases will have this
support in the Web UI and SNMP interfaces). When the user executes this command on WS1, WS1
creates a virtual session with the other controllers in the redundancy group (WS2, WS3 and WS4).
Once the virtual session is created, any command executed on WS1 is executed on the other
controllers at the same time. This is done by the cluster-protocol running on WS1, by duplicating
the commands and sending them to the group over the virtual connection:
Brocade Mobility RFS6000 and RFS7000 System Reference Guide
53-1001858-01
Reference time
Displays the time stamp at which the local clock was last set or corrected.
Clock Offset
Displays the time differential between controller time and the NTP resource.
Root delay
The total round-trip delay in seconds. This variable can take on both positive and
negative values, depending on the relative time and frequency offsets. The values
that normally appear in this field range from negative values of a few milliseconds
to positive values of several hundred milliseconds.
Root Dispersion
Displays the nominal error relative to the primary time source in seconds. The
values that normally appear in this field range from 0 to several hundred
milliseconds.
Configuring controller redundancy and clustering
5
293