Product Advantages

Advantage CDB for TDSQL MySQL
Cluster architecture

A more stable cluster architecture.

Non-cluster architecture

Only MySQL basic capability is available and failures occur easily when the standalone system is deployed.

Disaster recovery
Dual data centers (1 master/2 slaves)

Guarantees partial service in case of data center network failures

Single data center (1 master/1 slave)

Data center network failures will cause an interruption of services.

Synchronization mode
Forced synchronous replication

Host failures will not cause data loss/chaos.

Asynchronous/Semi-synchronous replication

Host failures will cause data loss/chaos.

Synchronization performance
Performance is 3 times that of semi-synchronization

Cross-data center forced synchronization under the same conditions reaches TPS = 9500

Low performance

Cross-data center synchronization under the same conditions reaches only TPS = 2200

Data reliability
3 hot and 3 cold backups

Data reliability of 99.99999%.

2 hot and 1 cold backups

Data reliability of 99.9996%.

Separated Read/write
Smart Loading

The CDB for TDSQL automatically selects the optimal-performance standby device for data reading.

Direct IP connection

Smart switching is not supported when standby device malfunctions.

Follow-up tracking
Rule filtering

Set rules to filter records

Rules cannot be configured

A third-party audit solution needs to be purchased.


Function Description

  • Security
    The overall protection architecture reaches or surpasses the standards for Class 3 information according to the protection standard.
    • Physical security: Provide exclusive zones and physical instances for customers in financial industries
    • Network security: Provides multiple security means at the network egress such as intrusion detection, vulnerability scanning, and login protection to guarantee network security. Data is transmitted inside VPC secret subnets and data is securely isolated.
    • Data security: File-level encryption capability allows for flexible configuration of encrypted content, tables, or database files. The performance consumption is only 3-5%, which is much lower than the application-layer encryption overhead.
    • Permission security: Supports permission control, covering different object granularities; including tables, views, functions, IP addresses and ports. Accounts of upper level administrators and high-risk configurations can be locked, and CLI operations can be controlled.
    • Follow-up tracking: Compliance audit can be conducted on database operations on a fine granularity basis, alerts are generated if risky behavior is detected. Audit logs are securely stored in accordance with national information security regulations.
  • Consistency and Availability
    Forced synchronization is used in combination with high availability to ensure complete consistency between the master and slave databases under the premise of 99.99% availability.
    • Forced synchronous replication: The Multi-thread Asynchronous Replication (MAR) is adopted between active and standby databases. After the standby database is synchronized with the active one, the active standby gives application transaction responses in an effort to ensure that data is not lost or incorrect. MAR greatly improves synchronization performance and delay in inter-data center synchronization is lower than 30 ms.
    • High availability scheduling: High Availability Scheduling (HAS) implements switching and scheduling inside database node groups, between multiple database node groups and between multiple network lines. The scheduling cluster is determined and the cluster election mechanism is configured to ensure complete disaster recovery. When an exception occurs, the access gateway cluster completes route switching within 500 ms, achieving availability as high as 99.99%.
    • Three data centers in two regions and disaster backup: Database node groups can be deployed in different data centers and architecture of three data centers in two regions covering the remote disaster backup node is supported. CDB for TDSQL provides flexible disaster backup modes for users. In addition to inter-data center deployment, CDB for TDSQL uses the Hadoop distributed file system (HDFS) to implement database disaster backup, with more than 6 copies of duplicated data, thereby achieving data reliability of up to 99.99999%.
    • Exception recovery: When a node malfunctions, CDB for TDSQL automatically re-creates a node within 30 minutes, without affecting normal services.
  • High Performance
    Hardware and architecture are comprehensively improved, making the performance 10 times better than your self-built databases.
    • PCI-e SSD: Enterprise-level solid state disks (SSD) are based on high-speed PCIExpress architecture and relevant controllers to provide hardware acceleration capabilities; Thread pool technology is also adopted to resolve restrictions on the maximum number of connections and system overhead caused by excessive threads.
    • Read/write splitting: Read/write splitting is supported, and the standby devices can be used to read data (not writable) for services, thereby greatly improving performance.
    • Performance optimization: CDB for TDSAQL monitors nodes, access, load and storage engines; and completely analyzes the 20 slowest query indicators to help users better troubleshoot database performance problems.


Standard Enterprise Finance Specific Distributed
Applicable scope

Applicable to all clients who have high requirements for availability and forced synchronization

Meets high requirements of governments and enterprises for database clouds, and provides secure and reliable cloud database services.

Meets finance-level rigorous database requirements and provides the highest-level service and performance for clients in the financial sector.

Provides database sharding capabilities and meets mass data storage requirements.

Purchase mode Online self-purchase Email Email Email
Service fee standard Billed as extreme services Customized service and billing standard Customized service and billing standard Customized service and billing standard
Price Price of the standard edition Price of the enterprise edition Price of the finance customization edition Price of the distribution edition
Basic configuration Node distributionGuangzhou and ShanghaiAny Tencent cloud data center in China

Any Tencent cloud data center in China//

Special finance area

Any Tencent cloud data center in China
Node architecture Two data centers in the same city/Single data center (configured as required) (configured as required) (configured as required)
HAS Supported Supported Supported Supported
MAR Supported Supported Supported Supported
HDFS Free of charge for 30 days due to release of the new product (configured as required) (configured as required) (configured as required)
Monitoring and log service Free of charge for 30 days due to release of the new product (configured as required) (configured as required) (configured as required)