Question

I need your help to optimize the query below. Let us assume we have a web application for articles. The software use two table;one is the article table and the second one is the users table. The article table hold the date when the article is created,the id,the body,the title & the section. Let us assume that we have one section called “news” and there are one million article belong to news section. So in this case, how to optimize the following query:

SELECT username,title FROM article,users 
WHERE article.auther_id=users.id AND section LIKE 'news' 
ORDER BY article.date DESC 
LIMIT 0,40

The table structures are:

CREATE TABLE `article` (
`id` INT NOT NULL AUTO_INCREMENT PRIMARY KEY ,
`title` VARCHAR( 100 ) NOT NULL ,
`body` VARCHAR( 200 ) NOT NULL ,
`date` VARCHAR( 30 ) NOT NULL ,
`auther_id` INT NOT NULL ,
`section` VARCHAR( 30 ) NOT NULL
) ENGINE = MYISAM ;


CREATE TABLE `users` (
`id` INT NOT NULL AUTO_INCREMENT PRIMARY KEY ,
`username` VARCHAR( 30 ) NOT NULL
) ENGINE = MYISAM ;

I tried to create one index that consists of the section & the date but it is not the best,because if we have 2 millions record and one million of them belong to one section,the DB will scan one million row.

Answer 1

You need to create an index on (section, date).

Don’t include auther_id as a leading column: articles will be leading in the join and no searching will be performed on this column.

Since there is a LIMIT 0, 40 in your query, MySQL will not have to scan the whole index. It will just pick the first 40 records.

Here’s a test script to check:

CREATE TABLE `article` (
`id` INT NOT NULL AUTO_INCREMENT PRIMARY KEY ,
`title` VARCHAR( 100 ) NOT NULL ,
`body` VARCHAR( 200 ) NOT NULL ,
`date` VARCHAR( 30 ) NOT NULL ,
`auther_id` INT NOT NULL ,
`section` VARCHAR( 30 ) NOT NULL
) ENGINE = MYISAM ;


CREATE TABLE `users` (
`id` INT NOT NULL AUTO_INCREMENT PRIMARY KEY ,
`username` VARCHAR( 30 ) NOT NULL
) ENGINE = MYISAM ;

INSERT
INTO    article
SELECT  id,
        CONCAT('Title ', id),
        CONCAT('Body ', id),
        DATE_FORMAT('2009-12-18' - INTERVAL id MINUTE, '%Y-%m-%d %H:%i:%S'),
        (id - 1) % 500 + 1,
        'news'
FROM    t_source;

INSERT
INTO    users
SELECT  id, CONCAT('Username ', id)
FROM    t_source
LIMIT 500;

CREATE INDEX ix_article_section_date ON article (section, date);

SELECT  username,title
FROM    article
JOIN    users
ON      users.id = article.auther_id
WHERE   section = 'news'
ORDER BY
        article.date DESC
LIMIT 0, 40;

t_source is a dummy table with 1,000,000 rows in it.

The final query completes in 0.0018 s on my machine (instantly)

Here’s the execution plan:

1, 'SIMPLE', 'article', 'range', 'ix_article_section_date', 'ix_article_section_date', '92', '', 999998, 'Using where'
1, 'SIMPLE', 'users', 'eq_ref', 'PRIMARY', 'PRIMARY', '4', 'test.article.auther_id', 1, ''