AnalyticDB for PostgreSQL, formerly known as HybridDB for PostgreSQL, is a massively parallel processing (MPP) data warehousing service designed to analyze large volumes of data online.

AnalyticDB for PostgreSQL is developed based on the open source Greenplum Database project and enhanced with in-depth extensions by Alibaba Cloud. It is compatible with ANSI SQL 2003 syntax and the PostgreSQL and Oracle database ecosystems. It also supports row-oriented storage and column-oriented storage. AnalyticDB for PostgreSQL processes petabytes of data online at a high performance and supports highly concurrent online queries. This makes it a competitive data warehousing solution in various industries.


  • Adaptable to variable workloads with no optimization required.

    AnalyticDB for PostgreSQL is fully compatible with SQL 2003 syntax and partially compatible with Oracle syntax. It also supports PL/SQL stored procedures. It offers new-generation query optimizers to relieve you from the need to optimize complex SQL statements.

  • Analyzes petabytes of data within seconds.

    AnalyticDB for PostgreSQL uses an MPP scale-out architecture to respond to queries for petabytes of data within seconds. It performs 10 times better than conventional database engines by using vectorized computing and intelligent columnstore indexing.

  • Provides high availability and always-on connectivity.

    AnalyticDB for PostgreSQL supports distributed transactions, ACID (short for atomicity, consistency, isolation, durability), redundancy for all nodes and data, plus automatic monitoring and failover.

  • Compatible with a wide variety of ecosystems.

    AnalyticDB for PostgreSQL supports mainstream business intelligence (BI) and extract, transform, load (ETL) tools. For example, it is integrated with the PostGIS extension to analyze geographic data and with the MADlib library to provide more than 300 built-in machine learning algorithms.

  • Enables data interconnection.

    AnalyticDB for PostgreSQL can synchronize data with various data sources by using tools such as Alibaba Cloud Data Transmission Service (DTS) and Dataworks. It also supports highly parallel access to OSS and Data Lake Analytics.


AnalyticDB for PostgreSQL uses the MPP architecture, which enables an instance to be composed of multiple compute nodes. You can add compute nodes to an instance to linearly scale out its storage capacity and maintain a stable response time. Each instance is composed of one coordinator node and multiple compute nodes.
  • Coordinator node
    • Receives query requests and determines distributed query plans.
  • Compute node
    • Provides massively parallel processing.
    • Stores data in dual copies on each partition.
    • Automatically backs up data to OSS on a regular basis.


Note Since August 23, 2019, the basic building block of an AnalyticDB for PostgreSQL instance has been changed from compute group to compute node. A compute group contains multiple partitions, whereas a compute node corresponds to an MPP partition. This simplifies the instance type definition and complies with cluster database naming conventions. For details about the mappings between compute nodes and compute groups, see Mappings between compute node types and compute group types.