PostgreSQL
Managed PostgreSQL with extensions, high availability, and enterprise-grade reliability.
Why teams use it
Managed PostgreSQL with extensions, high availability, and enterprise-grade reliability.
Managed PostgreSQL removes the installation, configuration, and patching overhead. You connect with any postgres-compatible driver, run your migrations, and query as normal.
Dedicated compute, ECC memory, NVMe-backed storage.
PostgreSQL runs behind a managed connection endpoint with PgBouncer support, dedicated compute, and storage sized separately from CPU and memory.
WAL archiving and backup targets stay managed by Tower, while your application keeps using standard PostgreSQL drivers and extensions.
What you get on Day One
The full PostgreSQL feature set on dedicated EPYC hardware with automatic operations.
PostgreSQL 14+
Latest stable versions with common extensions: PostGIS, pgvector, pg_cron, uuid-ossp, and more. Extension requests considered for new additions.
High availability
Synchronous streaming replication with automatic failover. Primary promotion on failure with no manual intervention required.
Point-in-time recovery
Continuous WAL archiving. Restore to any second within 30-day retention without relying solely on daily backups.
Connection pooling
PgBouncer included on all plans. Reduces connection overhead for high-concurrency applications without additional configuration.
Configuration & Platform Details
A concise reference for architecture reviews, security questionnaires, and capacity planning.
Platform specifications
- Engine
- PostgreSQL 17, 16, 15. Extension support includes pgvector, pg_repack, pg_audit, pg_cron, uuid-ossp, and standard contrib modules.
- Compute
- Dedicated vCPU and memory per instance. No shared resources with adjacent databases. Scales via SKU plans.
- Storage
- NVMe SSD-backed block storage. Data volume scales independently from compute, 1 GB to 1 TB. Separate WAL and backup repo volumes supported.
- Connection pooling
- PgBouncer included on all instances (3 replicas for HA). Transaction and session pooling modes configurable.
- High availability
- Streaming replication with 1–10 replicas. Automatic failover managed by the platform. Separate read-write (primary) and read-only (replica) endpoints. Pod anti-affinity spreads instances across nodes.
- Backups
- Full, differential, and incremental backups. Continuous WAL archiving for point-in-time recovery. Default schedule: weekly full backups. Supports local, S3-compatible, GCS, and Azure Blob storage destinations.
- Point-in-time recovery
- Restore to any second within the retention window (default 14 days). Supports restore to specific timestamp, backup ID, transaction ID, or timeline.
- Monitoring
- Built-in query analytics and performance monitoring enabled by default. Replication lag tracking and database health metrics included.
- Security
- TLS support available. SCRAM-SHA-256 authentication. Network policies for namespace-level isolation. Encrypted backup storage.
- Networking
- Public access via dedicated LoadBalancer IPs (separate read-write and read-only endpoints on port 5432). Private access via internal DNS.
Pricing
Loading SKUs...