DocsICP FilingConsole
Documentation

Overview

Updated at:
Copy as MD
Product
My Favorites

PolarDB for PostgreSQL offers cross-node parallel execution. This feature uses multiple compute nodes to execute SQL queries in a distributed manner. It accelerates analytical queries for PolarDB for PostgreSQL, fully utilizes the high I/O throughput capacity of the PolarFileSystem storage layer, and improves the CPU and memory resource utilization of all compute nodes.

Prerequisites

  • For clusters created before minor engine version V1.1.20 (released in January 2022), to use the cross-node parallel execution feature, go to the Quota Center and click Request in the Actions column for the PolarDB For PostgreSQL Cross-node Parallel Execution quota.
  • PolarDB for PostgreSQL cluster must be running minor engine version V1.1.11 or later. For more information about upgrading the minor engine version, see Version management.

Benefits

Cross-node parallel execution provides the following benefits:

  • Provides hybrid transactional and analytical processing (HTAP) capabilities:
    • You can execute analytical queries on real-time transactional processing (TP) data.
    • You can physically isolate read-only nodes for analytical queries from nodes for TP queries. This prevents any impact on your TP services.
  • Provides scalability with the decoupled storage and compute architecture of PolarDB for PostgreSQL:
    • If you need more computing power, you can add read-only nodes. The new nodes join the distributed parallel computing process without data resharding.
    • This design avoids data skew.

Scenarios

This feature is suitable for light analytical tasks in daily operations, such as account reconciliation.

Features

  • Cross-node parallel execution is primarily used to run analytical queries, which provides hybrid transactional and analytical processing (HTAP) capabilities. For more information, see Use cross-node parallel execution for analytical queries.

  • Cross-node parallel execution can also accelerate index building. For more information, see Use cross-node parallel execution to accelerate index creation.
  • You can use cross-node parallel execution to scan BRIN indexes. To enable this feature, set the polar_enable_px parameter to on. This lets you use BRIN indexes to further improve query performance.