Remote Client
A remote client acts as a node without actually being a member of the cluster. Instead of being directly connected to space, it is connected through a proxy - typically through a cache-agent.
A remote client does not contribute any of its resources towards maintaining the cluster. TIBCO BusinessEvents extends the same feature to allow its non-cache agents to connect to the cluster as remote members through a cache agent, that acts as a proxy. Using the remote client you can do better data management, as remote clients do not take part in cluster and thus dropping of one or more nodes from the cluster do not affect its processing.
Remote Client Behavior
- Only the non-cache (like inference) nodes can be remote clients.
- A remote client cannot contribute to cache storage. Thus, the Enable Cache Storage checkbox under the Processing Unit tab in CDD is ignored.
- Remote clients only operate as long as the cluster is up and reachable. A remote client cannot resume operations across cluster restart.
Best Practices
- For fault-tolerance of proxy nodes, open a remote listen port on two or more of your cache agents and specify a list of these nodes in remote discovery URLs. Thus, the chance of remote client losing connections to the cluster is minimized.
- To ensure consistency in deployment, either configure all inference nodes as remote client or none of them as remote. A mix of remote and non-remote connections is not recommended.
Copyright © TIBCO Software Inc. All rights reserved.