What is Managed PostgreSQL?

Managed PostgreSQL runs PostgreSQL databases for you on machines in your region, single node or as a three-node cluster with automatic failover.

Last updated 18 November 2026View as Markdown

A database is PostgreSQL 15, 16, 17 or 18 (17 by default) on machines in your region. A connection pooler runs beside PostgreSQL on its own port, and TLS is required on every connection by default. As a cluster it fails over automatically behind one address. You find it under Postgres in the Deutsche Cloud Console.

When to use it

  • An application needs a relational database and you do not want to run the machines, the failover, the backups or the certificates yourself.
  • You want to see what runs: every provisioning step, certificate action and storage change is written to the database's Activity, by the machine itself and by the platform.
  • You want a masked, writable copy of production data for a pull request, a preview or a migration rehearsal in seconds: see Branches.

Concepts

Term Meaning
Instance One database, identified by the name you give it. Single node or cluster.
Single node One machine. Its address is reserved before the machine exists and stays with the database if the machine is replaced.
Cluster Three machines with a load balancer in front. The direct and pooled ports go to the leader; failover moves the leader without changing the address. Needs an instance class with at least 8 GiB of memory.
Endpoint The IP address you connect to, with DNS names in its TLS certificate.
Direct port 5432: PostgreSQL itself.
Pooled port 6432: the connection pooler in transaction mode. Use this from applications.
Read port 6433: pooled connections to the replicas. Cluster only.
Reader An extra replica added to a running cluster for read capacity. Serves on the read ports with the other replicas; never becomes the leader.
Master password Generated at creation and filed in Secrets; the Connection tab reveals it to whoever may read that secret.
Activity The event stream on the database page: every step of provisioning, every certificate action, every storage change.
Branch A masked copy-on-write copy of the database that you can write to without touching the original.

More in Instances, clusters and endpoints.

Pricing

  • Nodes: every started hour of every node (a cluster is three, readers add one each), stopped ones included, at most the node's monthly price.
  • Storage: the allocated data volume of every node, per GB-month, measured daily.
  • Backups: backup storage up to the database's allocated storage is free; beyond it, the backup rate per GB-month (rds.backup.gb_month), measured daily.
  • Branches run on a branch pool machine built on first use; see Branches.
  • Parameter groups and subnet groups are configuration and are not charged.

The create form shows what the database costs a month before you create it.

Regions

Databases are in Germany (de1, zones de1a Falkenstein and de1b Nuremberg) unless your organization allows EU locations outside Germany; then they can also run in Helsinki, Finland (fi1). See Create a database.

Interfaces

  • The Deutsche Cloud Console, under Postgres.
  • The CLI: dcloud rds …; see the CLI Command Reference.
  • The HTTP API under /rds/api; see the API Reference.
  • Terraform: dcloud_db_instance, dcloud_db_read_replica, dcloud_db_parameters, dcloud_db_extension, dcloud_db_login, dcloud_db_replication_slot and dcloud_db_branch.
  • MCP: every operation is a tool named dcloud_rds_<resource>_<operation>.

Next steps

Did this page help you?