Unsupported Features and Limitations¶
Unsupported Features¶
Stored procedures
Triggers
Views
Events
User-defined functions
Foreign key reference and association
Full-text indexes and SPACE functions
Temporary tables
Compound statements such as BEGIN...END, LOOP...END LOOP, REPEAT...UNTIL...END REPEAT, and WHILE...DO...END WHILE
Process control statements such as IF and WHILE
RESET and FLUSH statements
BINLOG statement
HANDLER statement
INSTALL and UNINSTALL PLUGIN statements
Character sets other than ASCII, Latin1, binary, utf8, and utf8mb4
SYS schema
Optimizer Trace
X-Protocol
CHECKSUM TABLE syntax
Table maintenance statements, including ANALYZE, CHECK, CHECKSUM, OPTIMIZE, and REPAIR TABLE
Statements for assigning a value to or querying variable session
For example:
set @rowid=0;select @rowid:=@rowid+1,id from user;
SQL statements that use -- or /.../ to comment out a single line or multiple lines of code
DDM provides incomplete support for system variable queries. The returned values are variable values of RDS instances, instead of DDM kernel variable values. For example, the returned values of SELECT @@autocommit do not indicate the current transaction status.
Executing SET syntax to modify global variables
PARTITION syntax. Partitioned tables are not recommended.
LOAD XML statement
Unsupported Operators¶
Assignment operator (:=) is not supported.
Operator (->) is not supported. This operator can be executed successfully in a single table. An error is reported when this operator is executed in other types of tables.
Operator (->>) is not supported. This operator can be executed successfully in a single table. An error is reported when this operator is executed in other types of tables.
Expression IS UNKNOWN
Unsupported Functions¶
The compute layer of DDM does not support the following functions:
XML functions
Function ANY_VALUE()
Function ROW_COUNT()
Function COMPRESS()
Function SHA()
Function SHA1()
Function MD5()
Function AES_ENCRYPT()
Function AES_DECRYPT()
Aggregate function JSON_OBJECTAGG()
Aggregate function JSON_ARRAYAGG()
Aggregate function STD()
Aggregate function STDDEV()
Aggregate function STDDEV_POP()
Aggregate function STDDEV_SAMP()
Aggregate function VAR_POP()
Aggregate function VAR_SAMP()
Aggregate function VARIANCE()
SQL Syntax¶
SELECT
DISTINCTROW
Configuring options [HIGH_PRIORITY], [STRAIGHT_JOIN], [SQL_SMALL_RESULT], [SQL_BIG_RESULT], [SQL_BUFFER_RESULT], and [SQL_NO_CACHE] [SQL_CALC_FOUND_ROWS] in SELECT statements on DDM instances
SELECT ... GROUP BY ... WITH ROLLUP
SELECT ... ORDER BY ... WITH ROLLUP
WITH
Window functions
SELECT FOR UPDATE supports only simple queries and does not support statements such as JOIN, GROUP BY, ORDER BY, and LIMIT. Option [NOWAIT | SKIP LOCKED] for modifying FOR UPDATE is invalid for DDM.
DDM does not support multiple columns with the same name for each SELECT statement in UNION. Duplicate column names are used in the following SELECT statement:
SELECT id, id, name FROM t1 UNION SELECT pk, pk, name FROM t2;
SORT and LIMIT
LIMIT/OFFSET, value range: 0-2147483647
Aggregation
Function asc or desc cannot be used in the GROUP BY statement to sort out results.
Note
DDM automatically ignores keyword asc or desc after GROUP BY.
In MySQL versions earlier than 8.0.13, function asc or desc can be used in the GROUP BY statement to sort out results. In MySQL 8.0.13 or later, a syntax error is reported if you use function asc or desc this way. ORDER BY is recommended for sorting.
Subqueries
Subqueries that join grandparent queries are not supported.
Using subqueries in the HAVING clause and the JOIN ON condition is not supported.
Each derived table must have an alias.
A derived table cannot be a correlated subquery.
LOAD DATA
LOW_PRIORITY is not supported.
CONCURRENT is not supported.
PARTITION (partition_name [, partition_name] ...) is not supported.
LINES STARTING BY 'string' is not supported.
User-defined variables are not supported.
ESCAPED BY supports only '\\'.
If you have not specified a value for your auto-increment key when you insert a data record, DDM will not fill a value for the key. The auto-increment keys of data nodes of a DDM instance all take effect, so the auto-increment key values may be duplicate.
If the primary key or unique index is not routed to the same physical table, REPLACE does not take effect.
If the primary key or unique index is not routed to the same physical table, IGNORE does not take effect.
INSERT and REPLACE
INSERT DELAYED is not supported.
Only INSERT statements that contain sharding fields are supported.
PARTITION syntax is not supported. Partitioned tables are not recommended.
Setting YYYY of datetime (in the format of YYYY-MM-DD HH:MM:SS) to 1582 or any value smaller in INSERT statements is not supported.
Nesting a subquery in ON DUPLICATE KEY UPDATE of an INSERT statement is not supported. The following is an example:
INSERT INTO t1(a, b) SELECT * FROM(SELECT c, d FROM t2 UNION SELECT e, f FROM t3) AS dt ON DUPLICATE KEY UPDATE b = b + c;
Subquery c is used in the ON DUPLICATE KEY UPDATE clause.
The sharding key values in INSERT and REPLACE statements cannot be DEFAULT.
UPDATE and DELETE
Updating a sharding key value to DEFAULT is not supported.
Repeatedly updating the same field in one SQL statement is not supported.
Updating a sharding key using UPDATE JOIN is not supported. The following is an example:
UPDATE tbl_1 a, tbl_2 b set a.name=b.name where a.id=b.id;
name indicates the sharding key of table tbl_1.
Updating a sharding key by executing INSERT ON DUPLICATE KEY UPDATE is not supported.
Updating self-joins is not supported.
UPDATE tbl_1 a, tbl_1 b set a.tinyblob_col=concat(b.tinyblob_col, 'aaabbb');
UPDATE JOIN supports only joins with WHERE conditions.
The following is an example:
UPDATE tbl_3, tbl_4 SET tbl_3.varchar_col='dsgfdg';
Referencing other object columns in assignment statements or expressions is not supported when UPDATE JOIN syntax is used. The following is an example:
UPDATE tbl_1 a, tbl_2 b SET a.name=concat(b.name, 'aaaa'),b.name=concat(a.name, 'bbbb') ON a.id=b.id;
You can update a sharding field by two steps: delete the original sharding field and then insert a new field. During this process, the results of querying the sharding fields involved in the target table may be inconsistent.
DDL
SQL statements for modifying database names and sharding field names and types
SQL statements for creating and deleting schemas
Index FULL_TEXT
AS SELECT clause of the CREATE TABLE statement
CREATE TABLE ... LIKE statement
Dropping multiple tables with one SQL statement
Executing multiple SQL statements at the same time
Creating foreign keys for broadcast and sharded tables
Creating tables whose names are prefixed by _ddm
Creating temporary sharded or broadcast tables
Specifying globally unique keys in the CREATE TABLE statement
Indexes¶
Global secondary indexes
Global unique indexes. Unique keys and primary keys may not be globally unique.
Table Recycle Bins¶
Hints
Deleting tables by schema
Deleting tables by logical table
After a table is recovered, its globally unique sequence increases automatically but may not follow the last sequence value.
Shard configuration
Retaining copies with no time limit
Recovering data to a table with any name
Unlimited copies
Transactions¶
Savepoints
XA syntax. DDM has implemented distributed transactions through XA, so the user layer does not need to process the syntax.
Customizing the isolation level of a transaction. Currently, DDM supports only the READ COMMITTED isolation level. In consideration of compatibility, DDM does not report errors for any SQL statement (such as SET GLOBAL TRANSACTION ISOLATION LEVEL REPEATABLE READ) to set the database isolation level, but will ignore the modifications to the transaction isolation level.
Setting a transaction to read-only (START TRANSACTION READ ONLY). DDM can enable read/write of a transaction, instead of enabling read-only, to ensure compatibility.
Permissions¶
Column-level permissions
Subprogram-level permissions
Database Management Statements¶
SHOW TRIGGERS
Most of SHOW statements such as SHOW PROFILES, SHOW ERRORS, and SHOW WARNINGS
The following SHOW statements are randomly sent to a database shard. If database shards are on different RDS for MySQL instances, the returned variables or table information may be different.
SHOW TABLE STATUS;
SHOW VARIABLES Syntax;
SHOW WARNINGS Syntax does not support the combination of LIMIT and COUNT.
SHOW ERRORS Syntax does not support the combination of LIMIT and COUNT.
INFORMATION_SCHEMA¶
Only simple queries of SCHEMATA, TABLES, COLUMNS, STATISTICS, and PARTITIONS are supported. No subqueries, JOINs, aggregate functions, ORDER BY, and LIMIT are allowed.
Broadcast Tables¶
If a broadcast table is used, do not use any function that has different results returned each time it is executed. Otherwise, data inconsistency will occur between different shards. If such functions are indeed required, calculate their results, write the results to your SQL statements, and then execute the SQL statements on the broadcast table. Functions of this type include but are not limited to the following:
CONNECTION_ID()
CURDATE()
CURRENT_DATE()
CURRENT_TIME()
CURRENT_TIMESTAMP()
CURTIME()
LAST_INSERT_ID()
LOCALTIME()
LOCALTIMESTAMP()
NOW()
UNIX_TIMESTAMP()
UTC_DATE()
UTC_TIME()
UTC_TIMESTAMP()
CURRENT_ROLE()
CURRENT_USER()
FOUND_ROWS()
GET_LOCK()
IS_FREE_LOCK()
IS_USED_LOCK()
JSON_TABLE()
LOAD_FILE()
MASTER_POS_WAIT()
RAND()
RELEASE_ALL_LOCKS()
RELEASE_LOCK()
ROW_COUNT()
SESSION_USER()
SLEEP()
SYSDATE()
SYSTEM_USER()
USER()
UUID()
UUID_SHORT()