There are two options available as a communication method while setting up a TIBCO Spotfire Statistics Services (TSSS) cluster: Multicast and Messaging (JMS). All the nodes from the cluster will use the selected communication method among themselves. This option is available during Manager Node installation.
From the TIBCO Spotfire Statistics Services Installation and Administration:
• Multicast protocol communicates by sending short packets (datagrams), which can be received by all the nodes listening on a specific Multicast address and port. This protocol uses Java built-in support for object serialization to serialize and deserialize the messages sent using this protocol. This protocol provides a simple messaging bus that allows all nodes to stay in sync and distribute the workload with a very low overhead. However, the Multicast protocol uses UDP to transmit messages, and UDP packets can be dropped by switches and firewalls. To minimize packet loss, we recommend you use Multicast only if all the nodes of the cluster are physically located on the same subnet. Multicast protocol cannot be used for cloud computing.
• Messaging (JMS) is a very reliable messaging protocol, but it requires more resources from both manager and worker nodes to provide support for cluster messaging.
You must specify JMS Messaging if your server is running on a cloud. Otherwise, you should choose this protocol only if Multicast is unavailable or unreliable in a given networking context. Spotfire Statistics Services has its own built-in JMS support and does not require a separate JMS server product. Note that you cannot alter or substitute embedded JMS service that is based on Apache Active MQ.
If the TSSS cluster is configured with the Multicast protocol while installed on a cloud environment, then the Manager Node will not be able to communicate with the Worker Nodes. When viewing the Nodes URL in this state, the Worker Nodes will not be listed. Example Nodes URL (replace the <servername>, <port> and <servicename> as per your installation):
http://<servername>:<port>/<servicename>/api/v8/nodes
If all Manager and Worker nodes are not shown then they will not be able to communicate with each other.