Plan your deployment and install EDB Postgres AI for ClickHouse® on your infrastructure.
Deployment options
Your topology determines how data is stored, replicated, and served, as well as how much operational complexity you take on. For a full explanation of sharding, replication, and ClickHouse Keeper, see Architecture.
Single-node
ClickHouse runs as a standalone instance on a single host. No cluster coordination is needed. Suitable for development, testing, and single-tenant analytical workloads.
Multi-node cluster
Data is distributed across shards for horizontal scale and replicated for high availability. Replication requires ClickHouse Keeper running on at least three nodes. Keeper can run co-located on server nodes or on separate dedicated nodes.
Choosing an installation method
EDB provides three ways to install ClickHouse:
- Linux packages, for RHEL, Rocky Linux, and Debian
- Docker
- Kubernetes operator
If you're already running EDB Postgres AI for ClickHouse, see the upgrade section on the page for your installation method.