mysql query speed optimization

Question

I have the following which when run on its own is very quick, but when I am performing this for many entity_id the queries start to take longer and longer (the loop is a PHP foreach) for example this query only takes 0.078 but the same query on a different entity within the loop on takes upto 2.1 seconds, the queries seem to get slower and slower the more entities I put in the loop. Why is this? and how can I improve/optimize the query?

foreach($entity_ids as $entity_id) {
    SELECT COUNT(*) as prev, DATE_FORMAT(`created`, '%Y%m%d') AS date_group 
    FROM articles_entities 
    WHERE entity_id = '$entity_id' 
    AND `created` >= DATE_SUB(CURDATE(), INTERVAL 10 DAY) 
    GROUP BY date_group

    // store result
}

I have the following table structure:

CREATE TABLE `articles_entities` (
  `id` CHAR(36) NOT NULL,
  `article_id` CHAR(36) NOT NULL,
  `entity_id` CHAR(36) NOT NULL,
  `created` DATETIME DEFAULT NULL,
  `modified` DATETIME DEFAULT NULL,
  PRIMARY KEY (`id`),
  KEY `article_id` (`article_id`),
  KEY `entity_id` (`entity_id`),
  KEY `created` (`created`)
) ENGINE=MYISAM DEFAULT CHARSET=utf8;

Answer 1

It looks to me like you have an array of ID's, then want to pull records from your table (in a way controlled by your statement) where the ID field matches one of the array values.

Instead of using a LOOP in PHP to run multiple SQL statements, the best thing to do is build one master statement then use PHP to handle the results. You can accomplish this using the SQL IN statement:

    // where $entity_ids is an array eg 1,2,3,4,5

    $sql="SELECT entity_id AS 'alt_entity_id', COUNT(entity_id) as prev, DATE_FORMAT(`created`, '%Y%m%d') AS 'date_group'  
    FROM articles_entities 
    WHERE entity_id IN ".implode(",",$entity_ids)." 
    AND `created` >= DATE_SUB(CURDATE(), INTERVAL 10 DAY) 
    GROUP BY date_group, entity_id";
    // you may wish to revese the group fields, as you require, you may also wish to change the count field to date_group, depending on what you wish to be counted

This will run the original query one time, for all the id values you have, grouped by both date AND the id value passed. You can then use PHP to filter out the results for the specific id from the returned resultset.

This is far more efficient than the overhead produced by looping the execution of a query.

Your returned resultset will look something like:

 entity_id | count(entity_id) | date_group
 ----------|------------------|------------
     1     |        3         | 2010-04-01
     1     |        3         | 2010-03-01
     1     |        3         | 2010-02-01
     2     |        2         | 2010-01-01
     2     |        2         | 2010-02-01
     3     |        1         | 2010-06-01
     4     |        2         | 2010-06-01
     4     |        2         | 2010-02-01