> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tensor9.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Cluster
> Run multiple service adapter processes with shared coordination.
The **cluster** deployment runs multiple service-adapter processes behind one application-facing endpoint. The processes share PostgreSQL coordination state so they can make consistent decisions while scaling or restarting independently.
## When to use it
Choose a cluster when you need process-level redundancy, rolling restarts or more throughput than one binary can provide. The cluster does not change which API operations a mapping supports; it changes how the adapter is operated.
## What you operate
* Adapter processes and their rollout policy.
* The load-balancing or service-discovery layer in front of them.
* Shared PostgreSQL coordination state, including backup and recovery.
* Runtime credentials, destination-service access and network policy.
Keep coordination state separate from destination data in your inventory. A healthy process does not imply that either store is backed up or reachable.
## Rollouts
Use the local Tensor9 CLI to validate a configuration change, check compatibility and coordinate a rolling restart. Changes that require live migration between incompatible backends remain blocked until a migration path exists.