Movatterモバイル変換

Range columns partitioning is similar to range partitioning, but enables you to define partitions using ranges based on multiple column values. In addition, you can define the ranges using columns of types other than integer types.

RANGE COLUMNS partitioning differs significantly fromRANGE partitioning in the following ways:

The basic syntax for creating a table partitioned byRANGE COLUMNS is shown here:

CREATE TABLEtable_namePARTITION BY RANGE COLUMNS(column_list) (    PARTITIONpartition_name VALUES LESS THAN (value_list)[,    PARTITIONpartition_name VALUES LESS THAN (value_list)][,    ...])column_list:column_name[,column_name][, ...]value_list:value[,value][, ...]

In the syntax just shown,column_list is a list of one or more columns (sometimes called apartitioning column list), andvalue_list is a list of values (that is, it is apartition definition value list). Avalue_list must be supplied for each partition definition, and eachvalue_list must have the same number of values as thecolumn_list has columns. Generally speaking, if you useN columns in theCOLUMNS clause, then eachVALUES LESS THAN clause must also be supplied with a list ofN values.

The elements in the partitioning column list and in the value list defining each partition must occur in the same order. In addition, each element in the value list must be of the same data type as the corresponding element in the column list. However, the order of the column names in the partitioning column list and the value lists does not have to be the same as the order of the table column definitions in the main part of theCREATE TABLE statement. As with table partitioned byRANGE, you can useMAXVALUE to represent a value such that any legal value inserted into a given column is always less than this value. Here is an example of aCREATE TABLE statement that helps to illustrate all of these points:

mysql> CREATE TABLE rcx (    ->     a INT,    ->     b INT,    ->     c CHAR(3),    ->     d INT    -> )    -> PARTITION BY RANGE COLUMNS(a,d,c) (    ->     PARTITION p0 VALUES LESS THAN (5,10,'ggg'),    ->     PARTITION p1 VALUES LESS THAN (10,20,'mmm'),    ->     PARTITION p2 VALUES LESS THAN (15,30,'sss'),    ->     PARTITION p3 VALUES LESS THAN (MAXVALUE,MAXVALUE,MAXVALUE)    -> );Query OK, 0 rows affected (0.15 sec)

Tablercx contains the columnsa,b,c,d. The partitioning column list supplied to theCOLUMNS clause uses 3 of these columns, in the ordera,d,c. Each value list used to define a partition contains 3 values in the same order; that is, each value list tuple has the form (INT,INT,CHAR(3)), which corresponds to the data types used by columnsa,d, andc (in that order).

Placement of rows into partitions is determined by comparing the tuple from a row to be inserted that matches the column list in theCOLUMNS clause with the tuples used in theVALUES LESS THAN clauses to define partitions of the table. Because we are comparing tuples (that is, lists or sets of values) rather than scalar values, the semantics ofVALUES LESS THAN as used withRANGE COLUMNS partitions differs somewhat from the case with simpleRANGE partitions. InRANGE partitioning, a row generating an expression value that is equal to a limiting value in aVALUES LESS THAN is never placed in the corresponding partition; however, when usingRANGE COLUMNS partitioning, it is sometimes possible for a row whose partitioning column list's first element is equal in value to the that of the first element in aVALUES LESS THAN value list to be placed in the corresponding partition.

Consider theRANGE partitioned table created by this statement:

CREATE TABLE r1 (    a INT,    b INT)PARTITION BY RANGE (a)  (    PARTITION p0 VALUES LESS THAN (5),    PARTITION p1 VALUES LESS THAN (MAXVALUE));

If we insert 3 rows into this table such that the column value fora is5 for each row, all 3 rows are stored in partitionp1 because thea column value is in each case not less than 5, as we can see by executing the proper query against the Information SchemaPARTITIONS table:

mysql> INSERT INTO r1 VALUES (5,10), (5,11), (5,12);Query OK, 3 rows affected (0.00 sec)Records: 3  Duplicates: 0  Warnings: 0mysql> SELECT PARTITION_NAME, TABLE_ROWS    ->     FROM INFORMATION_SCHEMA.PARTITIONS    ->     WHERE TABLE_NAME = 'r1';+----------------+------------+| PARTITION_NAME | TABLE_ROWS |+----------------+------------+| p0             |          0 || p1             |          3 |+----------------+------------+2 rows in set (0.00 sec)

Now consider a similar tablerc1 that usesRANGE COLUMNS partitioning with both columnsa andb referenced in theCOLUMNS clause, created as shown here:

CREATE TABLE rc1 (    a INT,    b INT)PARTITION BY RANGE COLUMNS(a, b) (    PARTITION p0 VALUES LESS THAN (5, 12),    PARTITION p3 VALUES LESS THAN (MAXVALUE, MAXVALUE));

If we insert exactly the same rows intorc1 as we just inserted intor1, the distribution of the rows is quite different:

mysql> INSERT INTO rc1 VALUES (5,10), (5,11), (5,12);Query OK, 3 rows affected (0.00 sec)Records: 3  Duplicates: 0  Warnings: 0mysql> SELECT PARTITION_NAME, TABLE_ROWS    ->     FROM INFORMATION_SCHEMA.PARTITIONS    ->     WHERE TABLE_NAME = 'rc1';+----------------+------------+| PARTITION_NAME | TABLE_ROWS |+----------------+------------+| p0             |          2 || p3             |          1 |+----------------+------------+2 rows in set (0.00 sec)

This is because we are comparing rows rather than scalar values. We can compare the row values inserted with the limiting row value from theVALUES THAN LESS THAN clause used to define partitionp0 in tablerc1, like this:

mysql> SELECT (5,10) < (5,12), (5,11) < (5,12), (5,12) < (5,12);+-----------------+-----------------+-----------------+| (5,10) < (5,12) | (5,11) < (5,12) | (5,12) < (5,12) |+-----------------+-----------------+-----------------+|               1 |               1 |               0 |+-----------------+-----------------+-----------------+1 row in set (0.00 sec)

The 2 tuples(5,10) and(5,11) evaluate as less than(5,12), so they are stored in partitionp0. Since 5 is not less than 5 and 12 is not less than 12,(5,12) is considered not less than(5,12), and is stored in partitionp1.

TheSELECT statement in the preceding example could also have been written using explicit row constructors, like this:

SELECT ROW(5,10) < ROW(5,12), ROW(5,11) < ROW(5,12), ROW(5,12) < ROW(5,12);

For more information about the use of row constructors in MySQL, seeSection 15.2.15.5, “Row Subqueries”.

For a table partitioned byRANGE COLUMNS using only a single partitioning column, the storing of rows in partitions is the same as that of an equivalent table that is partitioned byRANGE. The followingCREATE TABLE statement creates a table partitioned byRANGE COLUMNS using 1 partitioning column:

CREATE TABLE rx (    a INT,    b INT)PARTITION BY RANGE COLUMNS (a)  (    PARTITION p0 VALUES LESS THAN (5),    PARTITION p1 VALUES LESS THAN (MAXVALUE));

If we insert the rows(5,10),(5,11), and(5,12) into this table, we can see that their placement is the same as it is for the tabler we created and populated earlier:

mysql> INSERT INTO rx VALUES (5,10), (5,11), (5,12);Query OK, 3 rows affected (0.00 sec)Records: 3  Duplicates: 0  Warnings: 0mysql> SELECT PARTITION_NAME,TABLE_ROWS    ->     FROM INFORMATION_SCHEMA.PARTITIONS    ->     WHERE TABLE_NAME = 'rx';+----------------+------------+| PARTITION_NAME | TABLE_ROWS |+----------------+------------+| p0             |          0 || p1             |          3 |+----------------+------------+2 rows in set (0.00 sec)

It is also possible to create tables partitioned byRANGE COLUMNS where limiting values for one or more columns are repeated in successive partition definitions. You can do this as long as the tuples of column values used to define the partitions are strictly increasing. For example, each of the followingCREATE TABLE statements is valid:

CREATE TABLE rc2 (    a INT,    b INT)PARTITION BY RANGE COLUMNS(a,b) (    PARTITION p0 VALUES LESS THAN (0,10),    PARTITION p1 VALUES LESS THAN (10,20),    PARTITION p2 VALUES LESS THAN (10,30),    PARTITION p3 VALUES LESS THAN (MAXVALUE,MAXVALUE) );CREATE TABLE rc3 (    a INT,    b INT)PARTITION BY RANGE COLUMNS(a,b) (    PARTITION p0 VALUES LESS THAN (0,10),    PARTITION p1 VALUES LESS THAN (10,20),    PARTITION p2 VALUES LESS THAN (10,30),    PARTITION p3 VALUES LESS THAN (10,35),    PARTITION p4 VALUES LESS THAN (20,40),    PARTITION p5 VALUES LESS THAN (MAXVALUE,MAXVALUE) );

The following statement also succeeds, even though it might appear at first glance that it would not, since the limiting value of columnb is 25 for partitionp0 and 20 for partitionp1, and the limiting value of columnc is 100 for partitionp1 and 50 for partitionp2:

CREATE TABLE rc4 (    a INT,    b INT,    c INT)PARTITION BY RANGE COLUMNS(a,b,c) (    PARTITION p0 VALUES LESS THAN (0,25,50),    PARTITION p1 VALUES LESS THAN (10,20,100),    PARTITION p2 VALUES LESS THAN (10,30,50),    PARTITION p3 VALUES LESS THAN (MAXVALUE,MAXVALUE,MAXVALUE) );

When designing tables partitioned byRANGE COLUMNS, you can always test successive partition definitions by comparing the desired tuples using themysql client, like this:

mysql> SELECT (0,25,50) < (10,20,100), (10,20,100) < (10,30,50);+-------------------------+--------------------------+| (0,25,50) < (10,20,100) | (10,20,100) < (10,30,50) |+-------------------------+--------------------------+|                       1 |                        1 |+-------------------------+--------------------------+1 row in set (0.00 sec)

If aCREATE TABLE statement contains partition definitions that are not in strictly increasing order, it fails with an error, as shown in this example:

mysql> CREATE TABLE rcf (    ->     a INT,    ->     b INT,    ->     c INT    -> )    -> PARTITION BY RANGE COLUMNS(a,b,c) (    ->     PARTITION p0 VALUES LESS THAN (0,25,50),    ->     PARTITION p1 VALUES LESS THAN (20,20,100),    ->     PARTITION p2 VALUES LESS THAN (10,30,50),    ->     PARTITION p3 VALUES LESS THAN (MAXVALUE,MAXVALUE,MAXVALUE)    ->  );ERROR 1493 (HY000): VALUES LESS THAN value must be strictly increasing for each partition

When you get such an error, you can deduce which partition definitions are invalid by making“less than” comparisons between their column lists. In this case, the problem is with the definition of partitionp2 because the tuple used to define it is not less than the tuple used to define partitionp3, as shown here:

mysql> SELECT (0,25,50) < (20,20,100), (20,20,100) < (10,30,50);+-------------------------+--------------------------+| (0,25,50) < (20,20,100) | (20,20,100) < (10,30,50) |+-------------------------+--------------------------+|                       1 |                        0 |+-------------------------+--------------------------+1 row in set (0.00 sec)

It is also possible forMAXVALUE to appear for the same column in more than oneVALUES LESS THAN clause when usingRANGE COLUMNS. However, the limiting values for individual columns in successive partition definitions should otherwise be increasing, there should be no more than one partition defined whereMAXVALUE is used as the upper limit for all column values, and this partition definition should appear last in the list ofPARTITION ... VALUES LESS THAN clauses. In addition, you cannot useMAXVALUE as the limiting value for the first column in more than one partition definition.

As stated previously, it is also possible withRANGE COLUMNS partitioning to use non-integer columns as partitioning columns. (SeeSection 26.2.3, “COLUMNS Partitioning”, for a complete listing of these.) Consider a table namedemployees (which is not partitioned), created using the following statement:

CREATE TABLE employees (    id INT NOT NULL,    fname VARCHAR(30),    lname VARCHAR(30),    hired DATE NOT NULL DEFAULT '1970-01-01',    separated DATE NOT NULL DEFAULT '9999-12-31',    job_code INT NOT NULL,    store_id INT NOT NULL);

UsingRANGE COLUMNS partitioning, you can create a version of this table that stores each row in one of four partitions based on the employee's last name, like this:

CREATE TABLE employees_by_lname (    id INT NOT NULL,    fname VARCHAR(30),    lname VARCHAR(30),    hired DATE NOT NULL DEFAULT '1970-01-01',    separated DATE NOT NULL DEFAULT '9999-12-31',    job_code INT NOT NULL,    store_id INT NOT NULL)PARTITION BY RANGE COLUMNS (lname)  (    PARTITION p0 VALUES LESS THAN ('g'),    PARTITION p1 VALUES LESS THAN ('m'),    PARTITION p2 VALUES LESS THAN ('t'),    PARTITION p3 VALUES LESS THAN (MAXVALUE));

Alternatively, you could cause theemployees table as created previously to be partitioned using this scheme by executing the followingALTER TABLE statement:

ALTER TABLE employees PARTITION BY RANGE COLUMNS (lname)  (    PARTITION p0 VALUES LESS THAN ('g'),    PARTITION p1 VALUES LESS THAN ('m'),    PARTITION p2 VALUES LESS THAN ('t'),    PARTITION p3 VALUES LESS THAN (MAXVALUE));

Similarly, you can cause theemployees table to be partitioned in such a way that each row is stored in one of several partitions based on the decade in which the corresponding employee was hired using theALTER TABLE statement shown here:

ALTER TABLE employees PARTITION BY RANGE COLUMNS (hired)  (    PARTITION p0 VALUES LESS THAN ('1970-01-01'),    PARTITION p1 VALUES LESS THAN ('1980-01-01'),    PARTITION p2 VALUES LESS THAN ('1990-01-01'),    PARTITION p3 VALUES LESS THAN ('2000-01-01'),    PARTITION p4 VALUES LESS THAN ('2010-01-01'),    PARTITION p5 VALUES LESS THAN (MAXVALUE));

SeeSection 15.1.20, “CREATE TABLE Statement”, for additional information aboutPARTITION BY RANGE COLUMNS syntax.