Understanding clustering

About clustering

Clusters are configured and managed using the Supervisor. A cluster is a group of blades on the same MSE chassis linked together to behave as a single unit that provides the combined port count of all the blades in the cluster. A larger port count provides flexibility: either conferences with more participants or several smaller conferences. You can configure two types of cluster:

MSE 8510 cluster: MSE Media2 blades running software version 4.1 or later support clustering. Currently up to three blades can be clustered, with one blade acting as a "master" and the other blades being "slaves" to this master.
Clustering provides you with the combined port count of the blades in the cluster. For example, on an MSE 8510 cluster of three blades each with 20 port licenses, the cluster can have either 30 HD ports or 15 HD+ ports and the master can allocate them to participants in conferences as necessary. This could be one large conference, or several smaller conferences. Note that in a cluster of MSE 8510 blades, SD ports are not available.
The maximum port counts for clusters comprising three MSE 8510 Media2 blades are as follows:
- For HD+ mode, the maximum number of port licenses that a three-blade cluster can utilize is 240. This will provide you with a total of 60 HD+ ports.
- For HD mode, the maximum number of port licenses that a three-blade cluster can utilize is 120. This will provide you with a total of 60 HD ports.
To configure media ports, on the MCU go to Settings > Media ports.
Telepresence Server 8710 cluster: Telepresence Server blades running software version 2 or later support clustering.
Clustering provides you with the combined video port count of the blades in the cluster. For example, on an MSE 8710 cluster of three blades each with 16 screen licenses, the cluster has 48 video ports and the master can allocate them to participants in conferences as necessary. This could be one large conference, or several smaller conferences.

Master blades

All of the port or screen licenses allocated to all the blades in a cluster are "inherited" by the master blade; all ports in the cluster are controlled by the master. Therefore, after you have configured a cluster, you must control conferences and MCU functionality through the master using either its web interface or through its API. All calls to the cluster are made through the master.

Slaves blades

Slave blades do not display the full MSE 8710 blade web interface. Only certain settings are available, such as network configuration, logging and upgrading. Similarly, a slave blade will only respond to a subset of API methods. For more information, refer to the relevant API documentation (available on www.tandberg.com).

General points

Some points to note about clustering:

Clustering requires the Cluster support feature key.
The MSE 8000 Supervisor blade must be running software version 2.1 or above to configure clustering.
You cannot mix the type of blades in a cluster but you can have both types of cluster in the same MSE chassis.
All blades in a cluster must be running the same version of software.
You assign the cluster roles (master/slave) to the slots in the chassis; if a blade fails, you can replace it with a blade of the same type and the cluster configuration will persist; however, what happens to active calls and conferences varies, as described below.
If you restart or remove the master, the slaves will also restart: all calls and conferences end.
Blades that do not support clustering can be installed into an MSE chassis alongside a cluster.
If the clustering configuration on the Supervisor and a blade disagree, then the Supervisor pushes the clustering configuration to the blade. (This might happen if you replace a slave blade with another blade of the same type.) The clustering configuration only includes clustering information; it does not configure network settings or anything else on the blade. If the Supervisor has pushed a configuration change to a blade, the Supervisor will prompt you to restart the blade.
Always keep a recent backup of the Supervisor.
If the Supervisor restarts or is removed, the cluster continues to function, conferences continue, and the cluster does not restart when the Supervisor reappears.
On the MCU, Call Detail Records (CDRs) are stored on and accessible on the master blade.
Slave blades only have admin logins.

Upgrading clustered blades

If you need to upgrade the blades in a cluster, first upload the new software images to each blade in the cluster and then restart the master. The slaves will automatically restart and the upgrade will be completed.