AnalyticDB for MySQL is available in two editions: Data Lakehouse Edition (V3.0) and Data Warehouse Edition (V3.0). This topic describes AnalyticDB for MySQL, its features, and the specifications of its different editions.
AnalyticDB for MySQL editions
Data Lakehouse Edition (V3.0)
Data Lakehouse Edition (V3.0) uses a decoupled architecture that separates computing from storage and integrates cost-effective batch processing and high-performance real-time analysis capabilities. Compared with Data Warehouse Edition, Data Lakehouse Edition (V3.0) features fully enhanced data collection, storage, computing, and application capabilities. This edition provides a graphical interface that allows you to configure data synchronization for Hudi on Object Storage Service (OSS) or C-Store tables in real time. The same copy of data stored in the underlying storage layer is used to perform both batch processing and real-time analysis. This prevents consistency and timeliness issues that may occur in the process of data synchronization. The computing layer supports the multi-language programmable Spark compute engine that uses standardized APIs. In addition, computing resources in Data Lakehouse Edition (V3.0) for batch processing and real-time analysis are physically isolated from each other. You can separately scale computing and storage resources based on your business requirements.
This edition is ideal for the following scenarios: data processing (such as data cleansing and standardization), multi-source aggregate analysis and table joining, and prediction and insights (such as machine learning and AI).
Elastic mode for Data Warehouse Edition (V3.0)
Data Warehouse Edition (V3.0) is built on a decoupled architecture that separates computing from storage, and can write large amounts of data in real time and perform high-performance real-time analysis. This edition allows you to individually scale up computing and storage resources based on your business requirements. It also provides tiered storage for hot and cold data, reducing storage costs. Futhermore, computing resources for batch processing and real-time analysis can be physically isolated.
This edition is ideal for writing large amounts of data in real time, performing complex extract-transform-load (ETL) operations, performing complex queries on large amounts of data, and analyzing historical data and logs.
- Basic Edition
The Basic Edition is deployed on a single node. It does not have the benefits of a distributed architecture. This edition supports tiered storage of hot and cold data. However, Alibaba Cloud does not provide an SLA guarantee for this edition. Furthermore, it does not support high availability, resource group isolation, or scheduled scaling. This edition is suitable for scenarios that do not require large amounts of data, high queries per second (QPS), or high availability. It is ideal for individual developers to run tests, and startups and small enterprises to handle basic business.
- Cluster Edition
The Cluster Edition is deployed across multiple nodes, and therefore delivers the benefits of a distributed architecture. This edition provides more powerful features to help enterprises with their development, testing, and production.
Reserved mode for Data Warehouse Edition (V3.0)
Data Warehouse Edition (V3.0) uses a decoupled architecture that separates computing from storage. It provides high throughput performance for real-time writes and concurrent queries, as well as quick response times. This edition is suitable for scenarios such as query acceleration, user portraits, interactive reports, and real-time data services.
Features
| Category | Feature | Data Lakehouse Edition (recommended) | Data Warehouse Edition | |
|---|---|---|---|---|
| Elastic mode | Reserved mode | |||
| Computing | XIHE analytical compute engine | Supported | Supported | Supported |
| Spark engine | Supported | Not supported | Not supported | |
| Storage | XUANWU analytical storage engine | Supported | Supported | Supported |
| Cost-effective Hudi storage | Supported | Not supported | Not supported | |
| Resource management | Resource group management | Supported | Supported for Cluster Edition only | Not supported |
| Scheduled scaling | Supported | Supported for Cluster Edition only | Not supported | |
| On-demand scaling | Supported | Not supported | Not supported | |
| Tiered storage of hot and cold data | N/A | Supported | Supported | Not supported |
| Data import | Real-time data import | Supported | Not supported | Not supported |
| Automatic metadata discovery | Supported | Not supported | Not supported | |
| Job development | SQL job development | Supported | Not supported | Not supported |
| Spark job development | Supported | Not supported | Not supported | |
| Job scheduling | N/A | Supported | Not supported Note This edition does not provide the native job scheduling capability. Job scheduling must be implemented by using Data Management (DMS) or DataWorks. | Not supported |
Specifications
Specifications of Data Lakehouse Edition (V3.0)
| Computing resources (in increments of 16 ACUs) | Storage resources (in increments of 24 ACUs) |
|---|---|
| Minimum: 16 AnalyticDB compute units (ACUs) Maximum: 2,048 ACUs | Minimum: 24 ACUs Maximum: 4,800 ACUs |
Specifications of elastic mode for Data Warehouse Edition (V3.0)
| Category | Computing resources |
|---|---|
| Basic Edition | 8 cores and 32 GB memory, 16 cores and 64 GB memory |
| Cluster Edition | Minimum: 16 cores and 64 GB memory Note Cluster Edition clusters that are purchased in the Chinese mainland as of September 1, 2022 can be deployed across multiple nodes if they have 16 cores and 64 GB memory or 24 cores and 96 GB memory. |
Specifications of reserved mode for Data Warehouse Edition (V3.0)
| Instance type | Specifications | ||
|---|---|---|---|
| CPU | Memory (GB) | Storage (GB) | |
| C8 | 24 cores | 192 | Minimum: 100 Maximum: 2,000 |
| C32 | 96 cores | 768 | Minimum: 100 Maximum: 8,000 |
FAQ
- How can I view the edition of my cluster?
You can log on to the AnalyticDB for MySQL console, go to the Cluster Information page of your cluster, and view the edition and mode of your cluster in the Cluster Attributes section.
- What are the differences between the elastic mode and reserved mode for Data Warehouse Edition (V3.0)?
- The billing methods of storage resources are different between the two modes.
- In reserved mode, you must specify the required storage capacity when you create the cluster. You are charged based on the specified storage capacity.
- In elastic mode, you do not need to specify the storage capacity when you create the cluster. You are charged based on the storage space that is actually occupied.
For example, if you purchase a pay-as-you-go cluster and the storage usage is 100 GB on a specific day, the storage resources that you must pay for are calculated based on the following formula: 100 GB × Usage duration. In elastic mode, AnalyticDB for MySQL imposes a minimum charge based on 20 GB of storage. You are charged for 20 GB of data storage even if the storage usage is less than 20 GB.
- In elastic mode, computing resources are separated from storage resources. Computing resources are used to process and compute data. Storage resources are used to read and write data. You can identify if performance bottlenecks of your workloads are caused by computing resources or storage resources, and scale your resources accordingly to reduce costs.
- The billing methods of storage resources are different between the two modes.
- In what scenarios may availability be affected?
Cluster availability may be affected if failures occur in a cluster or when the cluster undergoes configuration changes or version upgrades.