PolarDB-X integrates centralized and distributed architectures in one system. Start with a centralized deployment and scale out to distributed without data migration.
Overview
For most small and medium-sized enterprises (SMEs), a centralized database meets routine needs at lower cost and operational complexity. A distributed database handles high throughput, large-scale storage, and horizontal scaling, but costs more and requires greater technical expertise.
As business grows, an SME using a centralized database may need distributed scaling to handle higher concurrency and throughput.
PolarDB Distributed Edition (PolarDB-X) provides an integrated centralized-distributed architecture that combines distributed availability and scalability with centralized performance in a single system.
Features
In the integrated architecture, data nodes (DNs) run in centralized form, fully compatible with standalone MySQL. When you need distributed scaling, the architecture upgrades in place—distributed components connect to existing DNs without data migration or application changes.
Instance editions
PolarDB-X offers two editions: Standard Edition (centralized architecture) and Enterprise Edition (distributed architecture).
Standard Edition (centralized architecture)
PolarDB-X Standard Edition runs as a centralized deployment backed by multi-replica DNs. It supports specifications starting from 2 cores and 4 GB.
PolarDB-X Standard Edition uses the Paxos majority replication protocol for strong replica consistency and financial-grade high availability (RPO=0, RTO＜10 seconds). The self-developed Lizard distributed transaction engine provides more reliable high availability and delivers approximately 35% higher performance than the native MySQL distributed engine.
Enterprise Edition (distributed architecture)
PolarDB-X Enterprise Edition is a fully distributed deployment with compute nodes (CN), data nodes (DN), change data capture (CDC), columnar nodes (COLUMNAR), and global meta service (GMS). PolarDB-X Enterprise Edition is highly compatible with MySQL, and supports strongly consistent distributed transactions, parallel queries, and horizontal scaling.
Upgrade from Standard to Enterprise Edition
As business scales, PolarDB-X Standard Edition may hit centralized bottlenecks: large single tables degrading query performance, sustained high-concurrency loads, or unmet analytical requirements. At this point, vertical scaling alone becomes insufficient and cost-ineffective.
PolarDB-X allows you to upgrade an instance from Standard Edition to Enterprise Edition, leveraging distributed and HTAP capabilities while preserving the standalone MySQL experience.
Both editions share the same DNs. The upgrade adds distributed components (CN, CDC, GMS) without data migration. Only one copy of data exists, so rollback introduces no inconsistency.
Centralized tables convert in place to distributed single tables. Combined with Online DDL capabilities, this provides distributed scalability.
After upgrading an instance from Standard Edition to Enterprise Edition, the connection string stays the same and no application changes are needed. The upgrade causes only a minute-level transient connection interruption.
What is transparent distributed, deep compatibility with the centralized MySQL ecosystem, no application modifications required.
PolarDB-X uses table groups and partition groups to co-locate related data, optimizing distributed transactions and complex queries to match centralized performance in single-partition scenarios.
Storage resource pools and elastic specifications
PolarDB-X uses storage pools and Locality to enable on-demand distributed scaling.
Storage resource pools: DNs are grouped into non-overlapping pools. You can add or remove DNs at the pool level.
Locality: Associates database objects (databases, tables, partitions) with specific resource pools.
Two typical scenarios for on-demand distributed scaling:
Multi-tenant SaaS: After upgrading to Enterprise Edition, use vertical partitioning to distribute tenants across storage pools. Each pool maintains the single-table form for distributed scaling (see storage resource pool 1 in the diagram).
High-concurrency e-commerce: Use horizontal splitting to distribute data across multiple DNs in a resource pool (see storage resource pool 3 in the diagram).
Distribute centralized data using Online DDL:
Multiple single tables: Keep the single-table form and evolve into distributed vertical partitioning. After expanding DNs in a storage pool, tables distribute evenly across DNs.
Large tables: Convert online to distributed tables for horizontal scaling. After expanding DNs, partitions rebalance automatically.
Mixed workloads: Convert large tables to distributed tables while keeping single tables across multiple storage pools. This combines vertical and horizontal splitting for linear scaling through resource expansion.
Each DN can be independently scaled through data node management to match its actual resource needs and improve overall utilization.