PubSub+ Cache Redundancy and Network Deployments
This section discusses best practices for deploying PubSub+ Cache within a network with PubSub+ event brokers, including:
- Implementing PubSub+ Cache redundancy on a single event broker
- Using a Distributed Cache with a fault‑tolerant, redundant pair of event brokers
- Using a Distributed Cache in conjunction with software event brokers installed in primary and backup data-centers for datacenter redundancy
- Using Global Caching to extend the caching solution to a multi-region deployment
- Recommended Multiple-Node Routing topologies for these various network deployments of PubSub+ Cache
In general, fault-tolerant PubSub+ Cache deployments are achieved in an event mesh by ensuring every cache cluster contains multiple cache instances. Each instance connects to a different PubSub+ event broker within the mesh. This approach ensures that no single failure (be it network disruption, PubSub+ event broker service disruption, or failure of a PubSub+ Cache Instance) can affect the ability of PubSub+ Cache to continue both caching messages and servicing cache requests.
When setting up PubSub+ Cache within a network of event brokers, many deployment combinations and variations are possible. This section describes the most common deployment models. However, when planning a complex multi‑site deployment, we recommend engaging Solace Professional Services to assist with the network architecture design and to make recommendations on the best model for specific use cases.
An example deployment model is illustrated below. Any one (but only one) of the PubSub+ event brokers within the event mesh can be the Designated Router, and a failure affecting that event broker does not disrupt caching service for that cluster. The remaining event brokers and PubSub+ Cache Instances continue to function while connectivity to the Cache Manager is restored.
Publishers, subscribers, and, optionally, PubSub+ Cache Instances can use host-lists to identify their primary and secondary PubSub+ event brokers in the mesh to allow connections to be immediately re-established with the event mesh should the Designated Router fail or become unreachable.
