Release Notes
Product Announcements
Instance type | Description | Applicable Scenarios | |
Single-node instance | Single-node (formerly Basic Edition) instance | A single database node architecture is adopted, which is very cost-effective. Compute and storage are separated, and the underlying data is stored in three replicas in cloud disks. A basic edition node is deployed in a CVM instance, which has a higher database performance than self-built databases. Premium cloud disk is used as the underlying storage media, which is suitable for 90% I/O scenarios with a stable performance. | This instance type is suitable for personal learning, ISV software for small and medium-sized enterprises, web applications, and non-core small corporate systems. If a standalone instance fails, it takes a slightly longer time to recover than a CVM instance. |
Primary/Replica instance | Two-node (formerly HA Edition/Cluster Edition) instance with a one-primary-one-replica architecture. | SQL Server 2008 R2/2012/2014/2016 Enterprise or SQL Server 2012/2014/2016 Standard: It consists of one primary database and one mirror database. The classic one-primary-one-replica architecture is adopted, and all primary/replica instance nodes have the same specification. The primary and replica instances are deployed across racks/AZs. Each database corresponds to a monitoring agent that monitors the database through heartbeat in real time. Cross-AZ high availability is supported. In other words, the primary and replica instances can be deployed in different AZs. Object Storage: Offers data disaster recovery services, providing cold backup data. Tencent Cloud Management Cluster: Composed of independently deployed decision-making and configuration clusters, it serves as the management and scheduling center of the cluster, primarily overseeing the normal operation of database node groups, access gateway clusters, and object storage. | Replica instances improve the instance reliability. When a primary instance is created, a replica instance will be created at the same time, which will be invisible to users. When the primary instance fails, a primary-replica switch will be automatically triggered, and the database client will be disconnected momentarily. Therefore, the database client needs to support reconnection. The data sync between the primary and replica instances is async by default, so are the primary and read-only instances. |
| | SQL Server 2017/2019/2022 Enterprise: The Always On architecture is adopted, where you can add up to five read-only instances to create a cluster with higher availability, reliability, and scalability. The primary and replica instances are deployed across racks/AZs. Each database corresponds to a monitoring agent that monitors the database through heartbeat in real time. | This instance type is suitable for industry application scenarios, such as gaming, healthcare, medicine, internet, IoT, retail, ecommerce, logistics, insurance, securities, media, technical service, and automobile. |
| Multi-node instance with a one-primary-multiple-replica architecture (2 to 5 replica databases can be selected) | The supported versions are SQL Server 2017/2019/2022 Enterprise, with a one-primary-multiple-replica architecture. The replica nodes provide disaster recovery. After the read-only mode is enabled for replica nodes, they can also provide read-only addresses for read-only operations. The replica database can be deployed across AZs, making the business production environment more secure and capable of withstanding various unexpected situations, including overall IDC failures. | Suitable for business systems with high availability requirements, such as the financial industry and the e-commerce industry; suitable for business scenarios involving large-scale data processing, such as big data analysis and log processing. |
RO instance | Read-only instances | It is a single-node (with no replica) instance that supports read requests. It cannot exist independently; instead, it must be in a read-only group and bound to a two-node primary instance. A read-only group is a read-only instance group with the load balancing feature. | A read-only instance is standalone. When the physical server fails or a database replication exception occurs, it will take a long time (subject to the data volume) for the instance to recover. For business scenarios with a strong dependency on read-only requests, we recommend that you create multiple read-only instances to share the read pressure. |
フィードバック