Page History

Definitions

The topic is being broken up into three sections:

• Federation (a set of distributed exchanges and queues, seperately managed and wired together)
• Clustering (a set of distributed exchanges and queues managed as a single entity)
• HA / Fault tolarance (the ability for exchanges and queues, clustered, federated, or isolated to replicated important state to a backup (failover partner))
Failover (the ability for a client to reconnect to the failover partner)
• Custer for Reliability/Fault Tolarance
  • Cluster members replicate state.
  • If one cluster fails, clients can fail-over to another.
• Cluster for throughput
  • Distribute load to multiple brokers for higher throughput.

NOTE: clustering for reliability and throughput have very different goals.
They can be combined: e.g. distributing load across a set of reliable primary-backup pairs.

Requirements/Use Cases

• Reliability Requirements
• ClusteringAndFederation

Design notes

• Cluster Design Note - primary/backup pairing, transaction loggingCluster for reliability. A reliable broker cluster can participate as a single broker in federation or throughput clusters.
• Federation Design Note - Discussion of what has been done to date in C++
• Java Federation Design Proposal - Discussion of what could be implemented in the Java Broker.

Historical

• Old Clustering Design Note