How do I add indices to MySQL tables?

445
votes

I've got a very large MySQL table with about 150,000 rows of data. Currently, when I try and run

SELECT * FROM table WHERE id = '1';

the code runs fine as the ID field is the primary index. However, for a recent development in the project, I have to search the database by another field. For example:

SELECT * FROM table WHERE product_id = '1';

This field was not previously indexed; however, I've added one, so mysql now indexes the field, but when I try to run the above query, it runs very slowly. An EXPLAIN query reveals that there is no index for the product_id field when I've already added one, and as a result the query takes any where from 20 minutes to 30 minutes to return a single row.

My full EXPLAIN results are:

| id | select_type | table | type | possible_keys| key  | key_len | ref  | rows  | Extra       |
+----+-------------+-------+------+--------------+------+---------+------+-------+------------------+
|  1 | SIMPLE      | table | ALL  | NULL         | NULL | NULL    | NULL |157211 | Using where |
+----+-------------+-------+------+--------------+------+---------+------+-------+------------------+

It might be helpful to note that I've just taken a look, and ID field is stored as INT whereas the PRODUCT_ID field is stored as VARCHAR. Could this be the source of the problem?

7
mysqloptimizationindexingrow
Can you post the full EXPLAIN results? Are you certain it's that there's no index? Or is the index there, but MySQL's choosing not to use it? – VoteyDisciple
A large table would have 150,000,000 records. A very large table has 15,000,000,000 records. A table of average size has 150,000. For future reference. – usumoio
Be aware that 'OR' can make MySql not use indexes. I had a query with 3 OR's. Each mached an index, and ran in 15ms, All together took between 25sec and timeout. So I made 3 queries and UNION'ed them together, it also took 15ms on 500.000 rows. – Leif Neland
Consider the data type you are storing. Performance may change based on your data type you are comparing. As you said PRODUCT_ID is a VARCHAR data type, try changing it to a INT and index the column. – gilbertdim

7 Answers

668
votes
ALTER TABLE `table` ADD INDEX `product_id_index` (`product_id`)

Never compare integer to strings in MySQL. If id is int, remove the quotes.

160
votes
ALTER TABLE TABLE_NAME ADD INDEX (COLUMN_NAME);
89
votes

You can use this syntax to add an index and control the kind of index (HASH or BTREE).

create index your_index_name on your_table_name(your_column_name) using HASH;

or

create index your_index_name on your_table_name(your_column_name) using BTREE;

You can learn about differences between BTREE and HASH indexes here: http://dev.mysql.com/doc/refman/5.5/en/index-btree-hash.html

66
votes

Indexes of two types can be added: when you define a primary key, MySQL will take it as index by default.

Explanation

Primary key as index

Consider you have a tbl_student table and you want student_id as primary key:

ALTER TABLE `tbl_student` ADD PRIMARY KEY (`student_id`)

Above statement adds a primary key, which means that indexed values must be unique and cannot be NULL.

Specify index name

ALTER TABLE `tbl_student` ADD INDEX student_index (`student_id`)

Above statement will create an ordinary index with student_index name.

Create unique index

ALTER TABLE `tbl_student` ADD UNIQUE student_unique_index (`student_id`)

Here, student_unique_index is the index name assigned to student_id and creates an index for which values must be unique (here null can be accepted).

Fulltext option

ALTER TABLE `tbl_student` ADD FULLTEXT student_fulltext_index (`student_id`)

Above statement will create the Fulltext index name with student_fulltext_index, for which you need MyISAM Mysql Engine.

How to remove indexes ?

DROP INDEX `student_index` ON `tbl_student`

How to check available indexes?

SHOW INDEX FROM `tbl_student`
64
votes

It's worth noting that multiple field indexes can drastically improve your query performance. So in the above example we assume ProductID is the only field to lookup but were the query to say ProductID = 1 AND Category = 7 then a multiple column index helps. This is achieved with the following:

ALTER TABLE `table` ADD INDEX `index_name` (`col1`,`col2`)

Additionally the index should match the order of the query fields. In my extended example the index should be (ProductID,Category) not the other way around.

17
votes

You say you have an index, the explain says otherwise. However, if you really do, this is how to continue:

If you have an index on the column, and MySQL decides not to use it, it may by because:

  1. There's another index in the query MySQL deems more appropriate to use, and it can use only one. The solution is usually an index spanning multiple columns if their normal method of retrieval is by value of more then one column.
  2. MySQL decides there are to many matching rows, and thinks a tablescan is probably faster. If that isn't the case, sometimes an ANALYZE TABLE helps.
  3. In more complex queries, it decides not to use it based on extremely intelligent thought-out voodoo in the query-plan that for some reason just not fits your current requirements.

In the case of (2) or (3), you could coax MySQL into using the index by index hint sytax, but if you do, be sure run some tests to determine whether it actually improves performance to use the index as you hint it.

1
votes

A better option is to add the constraints directly during CREATE TABLE query (assuming you have the information about the tables)

CREATE TABLE products(
    productId INT AUTO_INCREMENT PRIMARY KEY,
    productName varchar(100) not null,
    categoryId INT NOT NULL,
    CONSTRAINT fk_category
    FOREIGN KEY (categoryId) 
    REFERENCES categories(categoryId)
        ON UPDATE CASCADE
        ON DELETE CASCADE
) ENGINE=INNODB;