PolarDB:Cross-node parallel execution

Prerequisites

• For clusters created before V1.1.20 (January 2022), enable this feature manually. Go to the Quota Center, find the PolarDB PG Cross-node Parallel Execution quota, and click Request in the Operation column.

• The minor engine version of PolarDB for PostgreSQL (Compatible with Oracle) must be V1.1.11 or later. To upgrade, see Version management.

Benefits

Cross-node parallel execution offers these benefits:

• Hybrid transactional and analytical processing (HTAP):

  • Run analytical queries on transactional (TP) data in real time.

  • Analytical read-only nodes can be physically isolated from TP read-only nodes, preventing analytical queries from affecting TP services.

• Scales with the decoupled storage-compute architecture ofPolarDB for PostgreSQL (Compatible with Oracle):

  • Add read-only nodes as needed. New nodes join the distributed parallel computing process without data resharding.

  • Prevents data skew.

Scenarios

Suited for lightweight analytical workloads in daily operations, such as account reconciliation.

Features

• Cross-node parallel execution runs analytical queries and provides HTAP capabilities.Use cross-node parallel execution for analytical queries.

• Cross-node parallel execution also accelerates index building, including B-tree and global B-tree indexes.Use cross-node parallel execution to accelerate index creation.

• BRIN index scans are supported. Set the polar_enable_px parameter to on to use BRIN indexes for improved query performance.BRIN indexes.

• Run analytical queries at the system, session, or query level.Use cross-node parallel execution for analytical queries.

• PolarDB for PostgreSQL (Compatible with Oracle) supports cross-node parallel execution on partitioned tables for improved performance.Use cross-node parallel execution on partitioned tables.