Topic

RDS for MySQL

GaussDB (for MySQL)

Architecture

Traditional primary/standby architecture. Data is synchronized between the primary and standby nodes using binlog.

Decoupled storage and compute architecture. Compute nodes share the same data and data does not need to be synchronized using binlog.

Performance

Hundreds of thousands of QPS, delivering three times the performance of the open-source MySQL in high concurrency.

Millions of QPS, delivering seven times more performance comparing with open-source MySQL. In complex queries, operations, such as column extraction, conditional filtering, and aggregation calculation, can be pushed down to the storage layer, improving the performance by dozens of times compared with traditional databases.

Scalability

Up to five read replicas can be added for an instance. The time required for adding read replicas depends on the data volume. Adding read replicas require additional storage.

The storage can grow as needed, with up to 4 TB for an instance.

Up to 15 read replicas can be added for an instance. Thanks to the shared storage, the time required for adding read replicas is not affected by the data volume. In addition, no additional storage is needed for read replica creation.

The storage grows as needed with up to 128 TB for an instance.

Availability

If the primary instance fails, the standby instance can be automatically promoted to the primary, with an RTO of less than 30s.

If the primary node fails, a read replica can be automatically promoted to the primary, with an RTO of less than 10s. It has lower latency because no data synchronization is required between the primary node and read replicas using binlog.