RDS Cluster Edition for ApsaraDB RDS for PostgreSQL is a high-availability (HA) architecture built on one primary node and multiple secondary nodes with compute-storage separation. Secondary nodes are readable, so you can scale read capacity without provisioning separate read-only instances — reducing costs by 40% compared with RDS High-availability Edition.
How it works
An RDS cluster consists of one primary node and one or more secondary nodes. All nodes share the same underlying storage layer (compute-storage separation).
| Endpoint | Connects to | When to use |
|---|---|---|
| Read/write endpoint | Primary node | Write operations, transactions that require the latest data |
| Read-only routing endpoint | Secondary nodes | Read-heavy workloads; the cluster distributes traffic across secondary nodes based on the read weights you configure. Free of charge. |
During a failover, the HA system detects secondary node faults at second-level granularity (improved from minute-level), and a secondary node recovers within 10 minutes in 99% of cases.
Edition comparison
The following table compares the three editions of ApsaraDB RDS for PostgreSQL.
| Feature | RDS Basic Edition | RDS High-availability Edition | RDS Cluster Edition |
|---|---|---|---|
| Number of nodes | 1 | 2 (1 primary + 1 secondary). Add read-only instances separately to scale reads. | 2–9 (default: 1 primary + 2 secondary; minimum: 1 primary + 1 secondary; maximum: 1 primary + 8 secondary) |
| Readable secondary nodes | N/A | Not supported | Supported |
| Endpoint types | Read/write endpoint | Read/write endpoint | Read/write endpoint + Read-only routing endpoint |
| Maximum unavailable nodes | 0 | 1 | n − 1 (n = total nodes in the cluster) |
| Number of zones | 1 | ≤ 2 | ≤ number of nodes |
Benefits
Lower cost
Secondary nodes in RDS Cluster Edition are readable — no need to provision separate read-only instances to scale read capacity. This linearly increases the read capability of the RDS cluster and reduces costs by 40% compared with RDS High-availability Edition.
The read-only routing endpoint is free of charge. Add multiple secondary nodes to the endpoint and assign read weights to balance load across them. For details, see View and change the endpoints and port numbers.
Flexible scaling
Add or delete nodes at any time without recreating the cluster. Deploy secondary nodes across multiple zones to meet your availability and latency requirements. Each node is independently monitored so you can view the status of each node in the cluster. For details, see Add nodes to or delete nodes from an RDS cluster.
Higher reliability
Cross-zone disaster recovery — Deploy secondary nodes in different zones. All secondary nodes are candidates for failover, unlike RDS High-availability Edition, which has only one standby instance.
Faster fault detection and recovery — The reconstructed HA system detects secondary node faults in seconds instead of minutes. Elastic Block Storage (EBS) single-digit-second backups cut data restore time from dozens of minutes to 1 minute, and secondary nodes recover within 10 minutes in 99% of cases.
Limitations
Version requirements
RDS Cluster Edition requires PostgreSQL 14 or later.
Feature limitations
Cross-zone migration is not supported.
When creating, cloning, upgrading the major engine version of, or recreating a cluster, you can configure up to two secondary nodes.
Upgrading the major engine version does not automatically update read-only routing endpoints. If you use the cutover method to upgrade, delete the read-only routing endpoints before upgrading, then recreate them after the upgrade completes.
You cannot change an instance from RDS Basic Edition or RDS High-availability Edition to RDS Cluster Edition.