What Database Protocols Does TDSQL Boundless Support?
TDSQL Boundless is compatible with the MySQL 8.0 protocol, allowing users to treat it as a MySQL 8.0 instance. However, certain operations are restricted. For details, see MySQL Compatibility. What Is the Maximum Supported Capacity of TDSQL Boundless?
The maximum supported capacity of TDSQL Boundless is theoretically unlimited. As your business requirements grow, you can scale the database capacity by adding more nodes to accommodate increasing data storage and processing demands. Currently, the public cloud supports the deployment of TDSQL Boundless instances containing dozens of nodes.
TDSQL Boundless also provides a visual interface for easily performing horizontal scaling operations. Furthermore, it features built-in automatic data migration and capacity balancing, which automatically adjusts data distribution across nodes. This ensures that system performance and storage efficiency are always optimized without requiring manual intervention.
Does TDSQL Boundless Require a ShardKey?
TDSQL Boundless does not require defining a shard key, and its table creation syntax remains consistent with native MySQL. The sharding mechanism of TDSQL Boundless is based on MySQL's native partition tables. In most cases, a primary Hash partition table is sufficient to meet requirements. Hash partitions are distributed across all data nodes, balancing the write pressure.
Does TDSQL Boundless Require Partition Tables?
In single-node deployments, partition tables are primarily used to enhance SQL performance through partition pruning and to periodically clean data by drop partition. In TDSQL Boundless distributed scenarios, the benefits of using partition tables also include leveraging the write capabilities of multiple nodes, which is particularly important for processing large volumes of data.
When large-scale data migration is performed, it is recommended to pre-convert large tables into Hash-based partition tables. This approach leverages the multi-node capability of TDSQL Boundless to accelerate the data import process.
If a single table is created without pre-partitioning, all write operations will be concentrated on a single data node during the initial data import phase, which may cause an I/O bottleneck. TDSQL Boundless provides the feature of automatic splitting and data migration. However, if the table is not partitioned initially, this process can be slow. Additionally, replica balancing during the splitting and migration period introduces extra I/O overhead.
Creating partition tables enables you to maximize the capabilities of the TDSQL Boundless distributed database. The cost of this adaptation is low, requiring only modifications to the table creation statement without any additional adaptation to the business code. For example, if your TDSQL Boundless instance contains 30 nodes, creating a primary Hash partition table with 30 partitions causes TDSQL Boundless to create a primary replica on each node, thereby achieving replica balancing. Concurrently, incremental business data is evenly distributed across all nodes, resulting in relatively balanced pressure on each node.
In summary, creating partition tables is a recommended best practice to fully leverage the distributed capabilities of TDSQL Boundless and avoid potential performance bottlenecks.