ClickHouse analytics
Connect, load event data, manage access and recover a full snapshot.
Choose an analytics workload
ClickHouse is a column-oriented SQL database for reporting, event streams, logs and large aggregations. Use PostgreSQL or MySQL for transactional application data. The starter uses a dedicated 2 vCPU / 4 GB RAM server. Choose only a plan that lists ClickHouse; administrators can use shared engine plans or separate ClickHouse plans.
Connect securely
Connections use HTTPS on port 8443 by default. You can select another unused port at checkout or change it in Network access. Plain HTTP and native TCP ports are disabled. Download the deployment CA, keep hostname verification enabled, and allow your application’s public IP before connecting. DataGrip uses the ClickHouse driver with an HTTPS JDBC URL and sslrootcert. The Connections tab includes a curl example and an @clickhouse/client Node.js example.
Create tables and ingest
Create a database in Databases & users, then open the browser workbench or connect your SQL client. A MergeTree table needs an ORDER BY key chosen around your common filters. Batch inserts rather than sending one row at a time. A 4 GB server suits smaller analytics workloads; large joins, concurrency and ingestion need more memory. Query memory and execution limits protect the server from individual runaway queries.
Database users
Create separate credentials per application and select its database. Read-only users can SELECT and inspect its objects. Write users can manage tables and ingest data in that database. Customers cannot create global users, grant privileges, alter server settings, read host files or configure external storage. Create users and databases through the portal so your plan limits are enforced.
Metrics and maintenance
Metrics show active queries, total queries, rows read and inserted, active MergeTree parts, background merges, connections and compressed table storage. Counters reset on server restart. Extra storage and server-size upgrades use the same reviewed price and capacity flow as other engines. A single-node upgrade or maintenance restart briefly interrupts connections. Engine packages are pinned to a tested LTS release; OS patching does not silently change database versions.
Full cloud backups
ClickHouse native backups capture database schemas and table data, together with database-user permissions, and are encrypted before upload to cloud storage. Automatic daily backups and manual backups share your storage allowance and retention policy. A restore creates a separate paid server and restores the selected full snapshot; the source remains unchanged. There is no point-in-time recovery or automatic failover for ClickHouse in this release. Backups of several tables are not a cross-table transaction: pause dependent ingestion if your application requires one consistent boundary.