Product editions

Updated at:

PolarDB-X offers two editions to match your scale and workload requirements: Standard Edition for teams running centralized deployments, and Enterprise Edition for distributed, high-concurrency production workloads. Both editions share the same MySQL-compatible foundation, and you can upgrade from Standard to Enterprise without migrating data or changing your application.

image

Standard Edition (Centralized)

Best suited for teams migrating from self-managed MySQL or RDS, or businesses in early growth stages that want room to scale later without re-architecting.

Built on a standalone architecture, the Standard Edition delivers 30%–40% better performance than open source MySQL while adding distributed optimizations: Paxos multi-replica for high availability and the Lizard transaction system for consistent transactions. It is 100% compatible with the MySQL ecosystem and supports finance-grade high availability (HA) with three-data-center disaster recovery within the same city and a recovery point objective (RPO) of 0.

When to choose Standard Edition:

  • Your business is in its early stages with low data volume and concurrency, but you expect to scale in the future.

  • You are migrating from a self-managed MySQL database or an RDS instance from another cloud provider and want better cost-effectiveness and availability.

  • You require finance-grade disaster recovery with an RPO of 0.

Limitations:

  • Bounded by single-node performance and capacity. The maximum specification is 90 cores.

  • Keep the number of records per table between 5 million and 50 million to avoid performance degradation on large tables.

Enterprise Edition (Distributed)

Designed for production workloads that have outgrown a standalone architecture — high-concurrency traffic, very large tables, or storage that exceeds single-node capacity.

The Enterprise Edition uses a distributed architecture where compute nodes (CNs), data nodes (DNs), and a global meta service collaborate to handle distributed transactions, parallel queries, and data sharding. The cluster scales transparently to up to 1,024 nodes and petabytes of data. Like the Standard Edition, it provides finance-grade HA with three-data-center disaster recovery and an RPO of 0. It is also highly compatible with the MySQL ecosystem and supports hybrid transactional and analytical processing (HTAP) workloads through column store read-only instances.

When to choose Enterprise Edition:

  • Your business faces high-concurrency bottlenecks, such as peak traffic in e-commerce transaction scenarios.

  • A single table has grown very large, or total storage has exceeded the capacity of a standalone database.

  • You want to replace an open source sharding solution to reduce operations and maintenance (O&M) overhead.

Limitations:

  • For small specifications (CPU ≤ 32 cores), the overhead of the distributed architecture may result in slightly lower performance than the Standard Edition.

Compare editions

Standard Edition (Centralized) Enterprise Edition (Distributed)
Architecture Standalone with distributed optimizations: Paxos multi-replica and the Lizard transaction system Distributed: CNs + DNs + global meta service
MySQL compatibility 100% compatible Highly compatible
Performance 30%–40% improvement over open source MySQL Transparent horizontal scaling to 1,024 nodes and petabytes of data
High availability Finance-grade HA; RPO of 0 Finance-grade HA; RPO of 0
Upgrade path In-place upgrade to Enterprise Edition —

Instance types

Each edition provides different instance types to match your workload requirements.

Edition Instance type Description Use cases
Standard Edition (Centralized) Primary instance The only read/write node in the cluster. Handles all write requests and most read requests. Required for all Standard Edition clusters.
Read-only instance Physically isolated from the primary instance. Synchronizes data using the Paxos protocol to offload read traffic and scale read capacity linearly. High-concurrency reads, data synchronization, and report analysis.
Enterprise Edition (Distributed) Primary instance The read/write entry point for the distributed cluster. Comprises multiple CNs and DNs to handle distributed transactions, parallel queries, and data sharding. Required for all Enterprise Edition clusters.
Read-only instance A distributed read-only replica, resource-isolated from the primary instance. Supports strong consistency reads for read/write splitting and high-concurrency read scenarios. Light report analysis, mixed workloads, and high-concurrency reads.
Column store read-only instance A read-only replica with columnar storage. Data is stored on Object Storage Service (OSS) for low cost and optimized for analytical processing (AP) queries. Report analysis, mixed workloads, data archiving, and risk control or auditing.