AnalyticDB for MySQL fully leverages the elasticity enabled by cloud computing. In addition to the reserved mode and elastic mode of Data Warehouse Edition (V3.0), AnalyticDB for MySQL released Data Lakehouse Edition (V3.0) that runs in serverless mode.

AnalyticDB for MySQL Data Warehouse Edition (V3.0) in reserved mode utilizes a Shared-Nothing architecture that provides high scalability and concurrency. Computing and storage resources are coupled and are related to the number of nodes. You can scale clusters in or out to adjust your resource consumption. However, you cannot flexibly specify the ratio of computing resources to storage resources. When you scale nodes, a large amount of data must be migrated. This can take a long period of time and affect system loads. Compute and storage resources cannot be separately adjusted, which leads to limited cost-efficiency.

AnalyticDB for MySQL Data Warehouse Edition (V3.0) in elastic mode uses a compute/storage-decoupled architecture to deliver a unified storage layer, an independently scalable compute layer, and high performance that is comparable to the reserved mode. After compute and storage resources are decoupled, you can separately scale them to control total costs. Data is not migrated when compute resources are being scaled. This delivers better elasticity than Data Warehouse Edition (V3.0) reserved mode. Compute resources in Data Warehouse Edition (V3.0) elastic mode are divided into several resource groups that are physically isolated. Scheduled scaling can be configured for each resource group. This way, queries from different business and scenarios can be distributed to different resource groups to reduce mutual impact, improving business stability.

AnalyticDB for MySQL Data Lakehouse Edition (V3.0) is in a compute/storage-decoupled architecture. It delivers a complete serverless solution on top of the features and benefits of Data Warehouse Edition (V3.0). You do not need to reserve storage or compute resources for your clusters. In Data Lakehouse Edition (V3.0), serverless resource groups are supported, where resources are dynamically allocated for each query and released after each query is executed. You are charged based on the actual amount of resources that you use. This delivers a smooth scaling experience and high cost-efficiency. Data Lakehouse Edition (V3.0) enhances the stability and fault tolerance for time-consuming jobs in scenarios such as batch data processing and ETL. It also allows access to a wide range of external data sources, increasing the variety of possible data lake scenarios.