This document introduces a new feature of TencentDB for MySQL based on the LibraDB engine: the read-only analysis engine.
Note:
The read-only analytics engine is currently available for free trial, with official commercialization and billing set to commence on May 19, 2025. For product pricing details, please refer to Billing Overview. Background
TencentDB for MySQL has been extensively optimized to support high concurrency, strong consistency, and enterprise-grade database features. Built on the TXSQL engine, it delivers high-performance online transaction processing (OLTP). However, beyond supporting high-QPS transactional workloads, many business systems also require data mining and analysis capabilities to enable better business decision-making, drive iterative innovation, and quickly adapt to changing market conditions.
Traditional databases, in order to support high-performance online transaction processing capabilities and ensure the stability of business queries, typically adopt row-based storage structures and the volcano execution model. However, these choices render them inefficient in handling analytical queries. Naturally, some businesses opt for a "traditional database + data warehouse" solution to accommodate mixed workloads involving both transactions and analytics. Yet, this approach imposes significant maintenance costs on customers, requiring them to develop their own ETL tools for transferring data from the database to the data warehouse. Moreover, it fails to adequately address the requirements for data timeliness and consistency.
To address this, TencentDB for MySQL now offers support for the new read-only analysis engine, providing users with efficient and real-time data analysis capabilities.
What Is the Read-Only Analysis Engine
The read-only analysis engine is a new feature of TencentDB for MySQL designed for real-time data analysis and complex query scenarios. Its pluggable engine architecture enables flexible and convenient creation and termination, while providing users with the capability to process massive volumes of data and perform efficient, real-time complex analysis.
Supported Regions and Availability Zones
For details on the regions and availability zones supported by the Read-Only Analytics Engine, please refer to Regions and AZs. Feature Strengths
High-performance analysis engine LibraDB
The LibraDB engine delivers high-performance complex query analysis, enabling your business analysis systems to efficiently extract valuable insights from massive databases in a timely manner. It supports analysis-oriented acceleration features such as a vectorized execution engine and massively parallel processing (MPP). Whether it is large-scale multi-table JOINs, data aggregation and sorting, or complex nested SQL queries, the LibraDB engine provides an exceptional performance experience.
Pluggable analysis engine
The LibraDB engine is compatible with MySQL protocols and syntax, allowing users to run complex queries directly in LibraDB without modifying their business logic. Users can enable the read-only analysis engine based on actual business needs, and disable it at any time when analytical acceleration is not required, helping to control costs effectively.
Real-time columnar data loading capability
With the built-in data synchronization component of the LibraDB engine, existing data in TencentDB for MySQL can be quickly loaded into the read-only analysis engine. After the initial data load, all subsequent changes made to the data in the read-write instance can be synchronized in real time, ensuring consistency between row-based and columnar data. In addition, to address the inefficiency of data changes in traditional columnar storage under high-concurrency data update and deletion scenarios, the LibraDB engine offers columnar storage capabilities optimized for high-concurrency data updates, enabling real-time synchronization and achieving zero-latency performance.
Targeted data loading capability
In traditional read-only instances, all data from the primary database should be fully synchronized to the secondary database. However, with the read-only analysis engine, it is possible to load only specified objects into the engine, rather than requiring full synchronization. Users can choose to load only those databases and tables that need acceleration through the analysis engine, or those with analytical value, enabling flexible control over the disk space used by the read-only analysis engine.
Ultra-high data compression ratio
Leveraging a columnar storage structure, the engine delivers ultra-high data scan performance while also achieving an average compression ratio of 4 to 5 times, significantly reducing storage costs.
Comprehensive cloud-hosted capability
No Ops is required for complex ETL logic or backend database management. With a fully managed product design, you gain an out-of-the-box experience for data analysis capabilities. Additionally, the monitoring feature provides a carefully curated set of core metrics, covering everything from TXSQL to the analysis engine, and from the linkage layer to the storage layer. This simplifies complexity and helps you quickly assess instance health using key metrics, offering effective optimization guidance for your business systems. You can also define custom alarm thresholds to proactively prevent potential exceptions.
Applicable Scenarios
The read-only analysis engine is designed to provide users with real-time, high-performance data analysis, helping to eliminate the complex Ops challenges of building custom ETL tools. With its built-in features, users can easily create data analysis instances with a single click, using them as a foundation for business decision-making and fully unlocking the value of their data.
Report analysis and real-time dashboards
For report systems designed for internal enterprise analysis and management, users can view the real-time operational status of online business systems. It also applies to data analysis tasks for business operations. In such scenarios, the SQL queries are complex and variable in pattern, requiring high throughput and involving large volumes of online data. The read-only analysis engine meets the real-time and high-performance requirements of these types of workloads.
User profiling and behavior analysis
In advertising and game operations scenarios, in-depth analysis of user behavior and user profiling is often required, with the results used to support real-time business decisions. These scenarios typically involve large volumes of data, require timely responses, and have high query QPS. By using the read-only analysis engine, users can quickly obtain the necessary data for analysis, enabling accurate insights into user behavior that serve as a decision-making foundation for precise business targeting.
Real-time data warehouse
The read-only analysis engine can also be used in scenarios such as order analysis during major e-commerce promotions, waybill analysis in the logistics industry, performance analysis and metric computation in the financial sector, live streaming quality analysis, ad delivery analysis, intelligent dashboards, and probe analysis, providing ultra-high performance for complex queries.
Big data reconciliation and batch computing
In certain online services, especially those involving financial transactions, periodic data aggregation and reconciliation are required. Performing batch reconciliation on traditional row-based data is often inefficient and resource-intensive, making it difficult to meet business expectations in a timely manner. By leveraging the high-concurrency computing capabilities of the read-only analysis engine, users can fulfill these business needs with significantly improved efficiency.