How to partition a MySQL table based on char column?

Question

Is it possible to partition based on char column?

After reviewing the MySQL 5.1 documentation it appears that only integer types can be used.

Is this correct? Or can I use some function to convert the char into an integer?

The char field in question contains a unique identifier.

Answer 1

Partitioning in MySQL 5.1 can only deal with integer columns ( Source ). You can only use a few partitioning functions on non-integer columns. For example:

CREATE TABLE ti (id INT, amount DECIMAL(7,2), tr_date DATE)
   ENGINE=INNODB
   PARTITION BY HASH( MONTH(tr_date) )
   PARTITIONS 6;

You can also use key partitioning in MySQL 5.1, as long as the primary key includes all the columns in the table's partitioning function:

CREATE TABLE k1 (
   id CHAR(3) NOT NULL PRIMARY KEY,
   value int
)
PARTITION BY KEY(id)
PARTITIONS 10;

On the other hand, in MySQL 5.5, you can use range column partitioning or list column partitioning on a wide variety of data types, including character-based columns.

List Columns Example:

CREATE TABLE expenses (
   expense_date DATE NOT NULL,
   category VARCHAR(30),
   amount DECIMAL (10,3)
);

ALTER TABLE expenses
PARTITION BY LIST COLUMNS (category)
(
   PARTITION p01 VALUES IN ('lodging', 'food'),
   PARTITION p02 VALUES IN ('flights', 'ground transportation'),
   PARTITION p03 VALUES IN ('leisure', 'customer entertainment'),
   PARTITION p04 VALUES IN ('communications'),
   PARTITION p05 VALUES IN ('fees')
);

Range Columns Example:

CREATE TABLE range_test (
   code CHAR(3),
   value INT
)
PARTITION BY RANGE COLUMNS(code) (
   PARTITION p0 VALUES LESS THAN ('MMM'),
   PARTITION p1 VALUES LESS THAN ('ZZZ')
);

Further reading:

• MySQL 5.1 Reference Manual :: Key partitioning
• A deep look at MySQL 5.5 partitioning enhancements
• MySQL 5.5 Reference Manual :: Range columns partitioning
• MySQL 5.5 Reference Manual :: List columns partitioning

Answer 2

What do you hope to gain by PARTITIONing? HASH and LIST are unlikely to provide any performance gain. It is rarely useful to partition a table with less than a million rows.

Other comments on partitioning: http://mysql.rjweb.org/doc.php/partitionmaint

5.6 and 5.7 have relaxed some of the restrictions mention in the other answers, but still partitioning is rarely better than indexing.