AnalyticDB for MySQL in elastic clusters utilizes an architecture that separates storage resources from computing resources.
Elastic I/O resources are bundled into units and sold as elastic I/O units (EIUs) to measure the storage performance of AnalyticDB for MySQL clusters. You can use EIUs to scale up storage resources.
- The edition of the cluster is Elastic mode for Cluster Edition.
- If the cluster is deployed in the China (Guangzhou), China (Shenzhen), China (Hangzhou), China (Shanghai), China (Qingdao), China (Beijing), or China (Zhangjiakou) region, the cluster is of the Cluster Edition and its minimum size of Computing Resources is 16 cores and 64 GB memory.
- If the cluster is deployed in another region, its minimum size of Computing Resources is 32 cores and 128 GB memory.
- Data import: You have purchased a small amount of computing resources. A large amount of data is written, which causes I/O bottlenecks on storage nodes.
- High-concurrency point query: You have purchased a small amount of computing resources. Point queries are concurrently performed, which causes CPU bottlenecks of storage nodes.
EIU-based storage performance metrics
The overall performance of EIUs is proportional to the number of EIUs. The following table describes the storage performance metrics of each EIU.
|Maximum hot data storage||4 TB|
|Maximum throughput||350 MB/s|
- When you purchase a cluster
When you purchase a cluster, you can estimate the number of required EIUs based on the storage performance metrics of EIUs. For more information, see the "EIU-based storage performance metrics" section of this topic. If you cannot estimate the required storage, you can purchase only one EIU first and scale up storage resources subsequently.
- When you scale up the storage resources of a cluster
When you scale up the storage resources of a cluster, you can increase the number of EIUs or change EIU specifications to enhance the data read and write capabilities of the cluster.