High Availability and Fault Tolerance - Clustered Operation

TIBCO ActiveMatrix BPM can be deployed to provide a high-availability, fault-tolerant configuration, using active-active clustering.

Note: This type of configuration requires the following:
  • an underlying high-availability machine and software architecture
  • a high-availability database system
  • external load balancing software for SOAP/HTTP connections (if required)
  • homogenous software configuration. Each member of the cluster must use the same TIBCO ActiveMatrix BPM logical node type.

The service virtualization capabilities provided by the TIBCO ActiveMatrix BPM runtime manage and route traffic between the nodes as required.

If a node, host or cluster member fails - whether that failure is caused by a hardware or software failure - messages are automatically re-routed to the remaining nodes in the cluster. Although execution capacity is reduced, service is not interrupted and no data or messages are lost. Outstanding transactions associated with the failed node are rolled back.

Note: HTTP clients (including browsers logged into Workspace/Openspace) communicate with the BPM Client logical node, so will be logged out if the web application dies. Any pageflows in progress in those clients will be lost, and clients will need to log in again.

To prevent this problem, an external load balancer can be used in front of HTTP clients to manage and route traffic from clients between BPM nodes.

TIBCO ActiveMatrix client applications (those running on the same TIBCO ActiveMatrix runtime) communicate with TIBCO ActiveMatrix BPM using service virtualization, so they are not affected.