AnalyticDB for MySQL uses the following terms across its documentation and console.
Region
A region is a geographical location where Alibaba Cloud data centers are deployed. Regions are typically named after the cities where the data centers reside. For example, the Malaysia (Kuala Lumpur) region has its data centers in Kuala Lumpur.
Zone
A zone is an isolated location within a region with its own independent power supply and network. Each region contains multiple zones. For example, the China (Beijing) region has twelve zones, including Beijing Zone A and Beijing Zone B.
Resources deployed within the same zone share the same network, which minimizes latency and improves communication speed between services.
Cluster
A cluster is the top-level deployment unit in AnalyticDB for MySQL. Clusters are physically isolated from each other. Each cluster contains one physical database and one or more logical databases.
Physical database
A physical database is the underlying storage representation of a database on physical devices, consisting of data files, log files, and index files.
A physical database is created automatically when you create an AnalyticDB for MySQL cluster. It is divided into multiple shards, which are distributed across storage node groups. Each shard is managed by a Raft group.
Logical database
A logical database is the organizational structure of a database that contains tables, views, and indexes. Logical databases are created by executing CREATE DATABASE statements. A single AnalyticDB for MySQL cluster supports up to 2,048 logical databases.
All references to "databases" in AnalyticDB for MySQL documentation refer to logical databases.
A shard is a horizontal slice of a table, created by splitting the table based on a distribution key. Shards are distributed across storage node groups, so each storage node group manages a portion of the data, which helps improve system scalability and performance.
Each shard is managed by an independent Raft group. Multiple replicas of a Raft group are deployed across storage nodes within the same storage node group, and the Raft protocol ensures data consistency across replicas.
Distribution of shards
The number of shards in a cluster does not change after you modify the cluster configuration. To query the current shard count, run:
SELECT COUNT(1) FROM information_schema.kepler_meta_shards;Partition
A partition is a subdivision of a shard, created by splitting the shard based on a partition key. In most cases, a date column is used as the partition key to support lifecycle management on data.
Table
AnalyticDB for MySQL supports two table types: partitioned tables and replicated tables. For syntax details, see CREATE TABLE.
Partitioned table
A partitioned table (also called a standard table) distributes data horizontally across all shards based on a distribution key. Each shard is then subdivided into partitions based on a partition key.
Key characteristics:
Scales with the distributed system for high query throughput
Supports up to hundreds of billions of rows
Partitioned tables are the primary table type for large-scale analytical workloads.
Replicated table
A replicated table stores a full copy of its data on every shard in the cluster. Changes to a replicated table are broadcast to all shards to maintain consistency, which affects write performance.
Key characteristics:
Each shard holds a complete copy of the data
Recommended maximum size: 20,000 rows
Avoid frequent inserts, updates, or deletes
Elastic I/O unit (EIU)
An elastic I/O unit (EIU) is the resource unit used to scale storage and I/O capacity for AnalyticDB for MySQL clusters in elastic mode for Cluster Edition. Each EIU is a storage node group consisting of three storage nodes, providing data storage, write, query, and scan capabilities.
Purchase EIUs when creating a cluster, and scale up or out to meet high-throughput write or high-volume scan requirements. For performance metrics, see Scale up or out elastic I/O resources.
AnalyticDB compute unit (ACU)
An AnalyticDB compute unit (ACU) is the resource unit for AnalyticDB for MySQL Data Lakehouse Edition. Computing and storage resources are bundled into ACUs. Each ACU is approximately 1 core and 4 GB of memory.
Resource group
A resource group is an isolated pool of computing resources within a cluster. Resource groups are physically isolated from each other, so workloads assigned to different resource groups do not compete for resources.
Resource groups are supported on:
AnalyticDB for MySQL Data Warehouse Edition clusters in elastic mode for Cluster Edition
AnalyticDB for MySQL Data Lakehouse Edition clusters
Assign different job types to separate resource groups to prevent resource contention. For example, run batch processing and real-time analytics in separate resource groups to avoid performance degradation.