ShardingSphere 4.x Sharding Concept Sharding

Sharding Key

The database field used in sharding refers to the key field in horizontal sharding of the database (table). For example, in last number modulo of order ID sharding, order ID is taken as the sharding key. The full route executed when there is no sharding field in SQL has a poor performance. Besides single sharding column, ShardingSphere also supports multiple sharding columns.

Sharding Algorithm

Data sharding can be achieved by sharding algorithms through =, >=, <=, >, <, BETWEEN and IN. They need to be implemented by developers themselves and can be highly flexible.

Currently, 4 kinds of sharding algorithms are available. Since the sharding algorithm and business achievement are closely related, it extracts all kinds of scenarios by sharding strategies, instead of providing built-in sharding algorithms. Therefore, it can provide higher abstraction and the interface for developers to implement sharding algorithm by themselves.

• Precise Sharding Algorithm

PreciseShardingAlgorithm is to process the sharding case in which single sharding keys = and IN are used; StandardShardingStrategy needs to be used together.

• Range Sharding Algorithm

RangeShardingAlgorithm is to process the sharding case in which single sharding key BETWEEN AND、>、<、>=、<= is used; StandardShardingStrategy needs to be used together.

• Complex Keys Sharding Algorithm

ComplexKeysShardingAlgorithm is to process the sharding case in which multiple sharding keys are used; ComplexShardingStrategy needs to be used together. It has a relatively complex logic that requires developers to deal by themselves.

• Hint Sharding Algorithm

HintShardingAlgorithm is to process the sharding case in which Hint is used; HintShardingStrategyneeds to be used together.

Sharding Strategy

It includes the sharding key and the sharding algorithm, and the latter one is extracted out for its independence. Only sharding key + sharding algorithm, i.e., the sharding strategy, can be used in sharding operation. For now, 5 kinds of sharding strategies are available.

• Standard Sharding Strategy

StandardShardingStrategy provides support for the sharding operation of =, >, <, >=, <=, IN and BETWEEN AND in SQL. StandardShardingStrategy only supports single sharding keys and provides two sharding algorithms of PreciseShardingAlgorithm and RangeShardingAlgorithm. PreciseShardingAlgorithm is compulsory and used to operate the sharding of = and IN. RangeShardingAlgorithm is optional and used to operate the sharding of BETWEEN AND, >, <, >=, <=. BETWEEN AND in SQL will operate by way of all data node route without the configuration of RangeShardingAlgorithm.

• Complex Sharding Strategy

ComplexShardingStrategy provides support for the sharding operation of =, >, <, >=, <=, IN and BETWEEN AND in SQL. ComplexShardingStrategy supports multiple sharding keys, but since their relationships are so complex that there is not too much encapsulation, the combination of sharding keys and sharding operators are in the algorithm interface and achieved by developers with the most flexibility.

• Inline Sharding Strategy

With Groovy expressions, InlineShardingStrategy provides single-key support for the sharding operation of = and IN in SQL. Simple sharding algorithms can be used through a simple configuration to avoid laborious Java code developments. For example, t_user_$->{u_id % 8} means table t_user is divided into 8 tables according to u_id, with table names from t_user_0 to t_user_7.

• Hint Sharding Strategy

HintShardingStrategy refers to the sharding strategy which get sharding values by hint rather than extracted from SQL.

• None sharding strategy

NoneShardingStrategy refers to the strategy with no sharding.

SQL Hint

In the case that the ShardingColumn is not decided by SQL but other external conditions, SQL Hint can be used flexibly to inject ShardingColumn. For example, in the internal system, databases are divided according to the staff’s ID, but this column does not exist in the database. SQL Hint can be used by two ways, Java API and SQL comment (to be finished).