MariaDB starting with 10.3
Sequences are being introduced in MariaDB 10.3. Details may still change before 10.3 is released as beta.
A sequence is an object that generates a sequence of numeric values, as specified by the CREATE SEQUENCE statement.
CREATE SEQUENCE will create a sequence that generates new values when called with NEXT VALUE FOR sequence_name. It's an alternative to AUTO INCREMENT when one wants to have more control of how the numbers are generated. As the SEQUENCE caches values (up to CACHE) it can in some cases be much faster than AUTO INCREMENT. Another benefit is that one can access the last value generated by all used sequences, which solves one of the limitations with LAST_INSERT_ID().
DROP SEQUENCE can be used to drop a sequence.
Using sequence objects
To get the next value from a sequence, use
NEXT VALUE FOR sequence
or in Oracle mode (SQL_MODE=ORACLE)
For retrieving the last value used by the current connection from a sequence use:
PREVIOUS VALUE FOR sequence_name
or in Oracle mode (SQL_MODE=ORACLE)
If one wants to use Sequences in a master-master setup or with Galera one
INCREMENT=0. This will tell the Sequence to use auto_increment_increment and auto_increment_offset to generate unique values for each server.
MariaDB 10.3 supports both ANSI SQL and Oracle syntax for sequences.
SEQUENCE is implemented as a special kind of table, it uses the same namespace as tables. The benefits are that sequences show up in SHOW TABLES, and one can also create a sequence with CREATE TABLE and drop it with DROP TABLE. One can SELECT from it as from any other table. This ensures that all old tools that work with tables should work with sequences.
Since sequence objects act as regular tables in many contexts, they will be affected by LOCK TABLES. This is not the case in other DBMS, such as Oracle, where LOCK TABLE does not affect sequences.
One of the goals with the Sequence implementation is that all old tools, such as mysqldump, should work unchanged, while still keeping the normal usage of sequence standard compatibly.
To make this possible,
sequence is currently implemented as a table with a few exclusive properties.
The special properties for sequence tables are:
- A sequence table has always one row.
- When one creates a sequence, either with CREATE TABLE or CREATE SEQUENCE, one row will be inserted.
- If one tries to insert into a sequence table, the single row will be updated. This allows mysqldump to work but also gives the additional benefit that one can change all properties of a sequence with a single insert. New applications should of course also use
can't be done on Sequence objects.
- Doing a select on the sequence shows the current state of the sequence, except the values that are reserved in the cache. The column
next_valueshows the next value not reserved by the cache.
- FLUSH TABLES will close the sequence and the next sequence number generated will be according to what's stored in the Sequence object. In effect, this will discard the cached values.
- A lot of normal table operations work on Sequence tables. See next section.
Table operations that work with Sequence tables
- SHOW CREATE TABLE sequence_name. This shows the table structure that is behind the
SEQUENCEincluding the field names that can be used with SELECT or even CREATE TABLE.
- CREATE TABLE sequence-structure ... SEQUENCE=1
- ALTER TABLE sequence RENAME TO sequence2
- RENAME TABLE sequence_name TO new_sequence_name
- DROP TABLE sequence_name. This is allowed mainly to get old tools like mysqldump to work with sequence tables.
- SHOW TABLES
Internally, sequence tables are created as a normal table without rollback (the InnoDB, Aria and MySAM engines support this), wrapped by a sequence engine object. This allowed us to create sequences with almost no performance impact for normal tables. (The cost is one 'if' per insert if the binary log is enabled).
Underlying table structure
The following example shows the table structure of sequences and how it can be used as a table. (Output of results are slightly edited to make them easier to read)
create sequence t1; show create sequence t1\G ***** 1. row ***** CREATE SEQUENCE `t1` start with 1 minvalue 1 maxvalue 9223372036854775806 increment by 1 cache 1000 nocycle ENGINE=InnoDB
show create table t1\G ***** 1. row ***** Create Table: CREATE TABLE `t1` ( `next_not_cached_value` bigint(21) NOT NULL, `minimum_value` bigint(21) NOT NULL, `maximum_value` bigint(21) NOT NULL, `start_value` bigint(21) NOT NULL COMMENT 'start value when sequences is created or value if RESTART is used', `increment` bigint(21) NOT NULL COMMENT 'increment value', `cache_size` bigint(21) unsigned NOT NULL, `cycle_option` tinyint(1) unsigned NOT NULL COMMENT '0 if no cycles are allowed, 1 if the sequence should begin a new cycle when maximum_value is passed', `cycle_count` bigint(21) NOT NULL COMMENT 'How many cycles have been done' ) ENGINE=InnoDB SEQUENCE=1
MariaDB [test]> select * from t1\G next_not_cached_value: 1 minimum_value: 1 maximum_value: 9223372036854775806 start_value: 1 increment: 1 cache_size: 1000 cycle_option: 0 cycle_count: 0
cycle_count column is incremented every time the sequence wraps around.
- Thanks to Jianwe Zhao from Aliyun for his work on SEQUENCE in AliSQL, which gave ideas and inspiration for this work.
- Thanks to Peter Gulutzan,who helped test and gave useful comments about the implementation.