Spec-Zone .ru
спецификации, руководства, описания, API

18.2.4. HASH Partitioning

18.2.4.1. LINEAR HASH Partitioning

Partitioning by HASH is used primarily to ensure an even distribution of data among a predetermined number of partitions. With range or list partitioning, you must specify explicitly into which partition a given column value or set of column values is to be stored; with hash partitioning, MySQL takes care of this for you, and you need only specify a column value or expression based on a column value to be hashed and the number of partitions into which the partitioned table is to be divided.

To partition a table using HASH partitioning, it is necessary to append to the CREATE TABLE statement a PARTITION BY HASH (expr) clause, where expr is an expression that returns an integer. This can simply be the name of a column whose type is one of MySQL's integer types. In addition, you will most likely want to follow this with a PARTITIONS num clause, where num is a positive integer representing the number of partitions into which the table is to be divided.

For example, the following statement creates a table that uses hashing on the store_id column and is divided into 4 partitions:

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,    store_id INT)PARTITION BY HASH(store_id)PARTITIONS 4;

If you do not include a PARTITIONS clause, the number of partitions defaults to 1.

Using the PARTITIONS keyword without a number following it results in a syntax error.

You can also use an SQL expression that returns an integer for expr. For instance, you might want to partition based on the year in which an employee was hired. This can be done as shown here:

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,    store_id INT)PARTITION BY HASH( YEAR(hired) )PARTITIONS 4;

expr must return a nonconstant, nonrandom integer value (in other words, it should be varying but deterministic), and must not contain any prohibited constructs as described in Section 18.6, "Restrictions and Limitations on Partitioning". You should also keep in mind that this expression is evaluated each time a row is inserted or updated (or possibly deleted); this means that very complex expressions may give rise to performance issues, particularly when performing operations (such as batch inserts) that affect a great many rows at one time.

The most efficient hashing function is one which operates upon a single table column and whose value increases or decreases consistently with the column value, as this allows for "pruning" on ranges of partitions. That is, the more closely that the expression varies with the value of the column on which it is based, the more efficiently MySQL can use the expression for hash partitioning.

For example, where date_col is a column of type DATE, then the expression TO_DAYS(date_col) is said to vary directly with the value of date_col, because for every change in the value of date_col, the value of the expression changes in a consistent manner. The variance of the expression YEAR(date_col) with respect to date_col is not quite as direct as that of TO_DAYS(date_col), because not every possible change in date_col produces an equivalent change in YEAR(date_col). Even so, YEAR(date_col) is a good candidate for a hashing function, because it varies directly with a portion of date_col and there is no possible change in date_col that produces a disproportionate change in YEAR(date_col).

By way of contrast, suppose that you have a column named int_col whose type is INT. Now consider the expression POW(5-int_col,3) + 6. This would be a poor choice for a hashing function because a change in the value of int_col is not guaranteed to produce a proportional change in the value of the expression. Changing the value of int_col by a given amount can produce by widely different changes in the value of the expression. For example, changing int_col from 5 to 6 produces a change of -1 in the value of the expression, but changing the value of int_col from 6 to 7 produces a change of -7 in the expression value.

In other words, the more closely the graph of the column value versus the value of the expression follows a straight line as traced by the equation y=cx where c is some nonzero constant, the better the expression is suited to hashing. This has to do with the fact that the more nonlinear an expression is, the more uneven the distribution of data among the partitions it tends to produce.

In theory, pruning is also possible for expressions involving more than one column value, but determining which of such expressions are suitable can be quite difficult and time-consuming. For this reason, the use of hashing expressions involving multiple columns is not particularly recommended.

When PARTITION BY HASH is used, MySQL determines which partition of num partitions to use based on the modulus of the result of the user function. In other words, for an expression expr, the partition in which the record is stored is partition number N, where N = MOD(expr, num). Suppose that table t1 is defined as follows, so that it has 4 partitions:

CREATE TABLE t1 (col1 INT, col2 CHAR(5), col3 DATE)    PARTITION BY HASH( YEAR(col3) )    PARTITIONS 4;

If you insert a record into t1 whose col3 value is '2005-09-15', then the partition in which it is stored is determined as follows:

MOD(YEAR('2005-09-01'),4)=  MOD(2005,4)=  1

MySQL 5.6 also supports a variant of HASH partitioning known as linear hashing which employs a more complex algorithm for determining the placement of new rows inserted into the partitioned table. See Section 18.2.4.1, "LINEAR HASH Partitioning", for a description of this algorithm.

The user function is evaluated each time a record is inserted or updated. It may also—depending on the circumstances—be evaluated when records are deleted.

Note

If a table to be partitioned has a UNIQUE key, then any columns supplied as arguments to the HASH user function or to the KEY's column_list must be part of that key.