configuration work. Java codes are not helpful in the unified management of common configurations. Writing sharding algorithms with inline expressions, users can store rules together, making them easier alone or with sharding table and database; • Primary nodes need to be used for both reading and writing in the transaction; • Forcible primary database route based on SQL Hint; Unsupported • Data replication inconsistent; In the readwrite‐ splitting model, primary nodes need to be used for both reading and writing in the transaction. 4.6 HA 4.6.1 Background High availability is the most basic requirement of

configuration work. Java codes are not helpful in the unified management of common configurations. Writing sharding algorithms with inline expressions, users can store rules together, making them easier alone or with sharding table and database; • Primary nodes need to be used for both reading and writing in the transaction; • Forcible primary database route based on SQL Hint; Unsupported • Data replication inconsistent; In the readwrite‐ splitting model, primary nodes need to be used for both reading and writing in the transaction. 4.5 Governance 4.5.1 Background As the scale of data continues to expand,

configuration work. Java codes are not helpful in the unified management of common configurations. Writing sharding algorithms with inline expressions, users can store rules together, making them easier alone or with sharding table and database; • Primary nodes need to be used for both reading and writing in the transaction; • Forcible primary database route based on SQL Hint; Unsupported • Data replication inconsistent; In the readwrite‐ splitting model, primary nodes need to be used for both reading and writing in the transaction. 4.6 HA 4.6.1 Background High availability is the most basic requirement of

algorithm, using Java code implementation does not help to manage the configuration uniformly. But by writing the sharding algorithm through line expressions, the rule con‐ figuration can be effectively stored If there is no specific statement, the following implementations all use YAML 1.1 as the YAML writing specification. This does not prevent custom implementations of org.apache. shardingsphere.infra for data sharding, which allows users to create ShardingSphereData‐ Source objects directly by writing Java code, is flexible enough to integrate various types of business systems without relying on

algorithm, using Java code implementation does not help to manage the configuration uniformly. But by writing the sharding algorithm through line expressions, the rule con‐ figuration can be effectively stored for data sharding, which allows users to create ShardingSphereData‐ Source objects directly by writing Java code, is flexible enough to integrate various types of business systems without relying on Java API rule configuration allows users to directly create ShardingSphereData‐ Source objects by writing java code. The Java API configuration method is very flexible and can inte‐ grate various types

configuration work. Java codes are not helpful in the unified management of common configurations. Writing sharding algorithms with inline expressions, users can stored rules together, making them easier batch inserted SQL, if the inserted data crosses sharding, the user needs to rewrite SQL to avoid writing excessive data into the database. The differences between insert operation and query operation are: inconsistent. In the replica query model, the primary nodes need to be used for both reading and writing in the transaction. 3.4 Governance 3.4.1 Background As the scale of data continues to expand,

algorithm, using Java code implementation does not help to manage the configuration uniformly. But by writing the sharding algorithm through line expressions, the rule con‐ figuration can be effectively stored for data sharding, which allows users to create ShardingSphereData‐ Source objects directly by writing Java code, is flexible enough to integrate various types of business systems without relying on Java API rule configuration allows users to directly create ShardingSphereData‐ Source objects by writing java code. The Java API configuration method is very flexible and can inte‐ grate various types

algorithm, using Java code implementation does not help to manage the configuration uniformly. But by writing the sharding algorithm through line expressions, the rule con‐ figuration can be effectively stored for data sharding, which allows users to create ShardingSphereData‐ Source objects directly by writing Java code, is flexible enough to integrate various types of business systems without relying on Java API rule configuration allows users to directly create ShardingSphereData‐ Source objects by writing java code. The Java API configuration method is very flexible and can inte‐ grate various types

数据源停写， xx：任务 id STOP SCALING SOURCE WRITING 1234 RESTORE SCALING SOURCE WRITING jobId 旧的 ShardingSphere 数据源恢复写， xx：任务 id RESTORE SCALING SOURCE WRITING 1234 APPLY SCALING jobId 切换至新的 ShardingSphere EXECUTE_INCREMENTAL_TASK，全量迁移已完成，在增量迁移阶段。 选择一个业务低峰期，对源端库或数据操作入口做停写。 proxy 停写： mysql> STOP SCALING SOURCE WRITING 0130317c30317c3054317c7363616c696e675f6462; Query OK, 0 rows affected (0.07 sec) 数据一致性校验： mysql>

Application.This is discussed clearly in later chapters. mapper XML file prevents the burden of writing SQL statements repeatedly in the application. In comparison to JDBC, almost 95% of the code is reduced