PolarDB for PostgreSQL (Compatible with Oracle) is a cloud-native database service built on a compute-storage separated architecture. All compute nodes share a unified storage layer, enabling minute-level resource scaling, second-level fault recovery, and global data consistency. The service combines the stability, reliability, high performance, and scalability of commercial databases with the simplicity, openness, and rapid iteration of open source cloud databases. It is highly compatible with Oracle syntax, making it a practical migration target for Oracle workloads.
Key capabilities:
| Capability | Details |
|---|---|
| Elastic scaling | Scale compute resources in minutes to handle traffic spikes |
| Fast fault recovery | Recover from node failures in seconds |
| Built-in backup | Data backup and disaster recovery included at no extra cost |
| Massive read capacity | Up to 15 read-only nodes per Enterprise Edition cluster |
| Oracle compatibility | Run Oracle workloads without rewriting syntax |
Architecture
PolarDB for PostgreSQL (Compatible with Oracle) separates compute from storage. Multiple compute nodes share a single distributed storage layer, so:
Adding or removing read-only nodes does not require data migration or replication.
The primary node and all read-only nodes always read from the same storage, eliminating data synchronization lag.
Storage scales independently of compute resources.
Key concepts
| Term | Definition |
|---|---|
| Region | The geographical area where a PolarDB cluster is deployed. |
| Zone | A physically separate area within a region, with independent power and networking infrastructure. Instances in the same zone have lower network latency than those in different zones. See Alibaba Cloud's global infrastructure. |
| Cluster | One primary node plus multiple read-only nodes. Clusters support cross-zone deployments but cannot span multiple regions. Each cluster has a unique ID with the prefix pc. |
| Node | A physical compute unit within a cluster. Each node has a unique ID with the prefix pi. Nodes of the same type share the same specifications. |
| Primary node | The single read/write node in a cluster. Handles both reads and writes. Also called the read-write node. |
| Read-only node | A node that handles read traffic only. Enterprise Edition clusters support up to 15 read-only nodes. |
Console
Use the PolarDB console to create, connect to, and configure PolarDB clusters.