Question

So I was on tech support with our host provider about slow database response that I’d been experiencing on a crucial search page and after some investigating he told me that the query that I’m using was checking 67,998,176 rows. Now the largest table that appears on this query has at most 116 rows and the rest of them have an average of 25 rows each.

Here’s the SQL of my query

 SELECT COUNT( DISTINCT `A`.`id`) AS `total_num_resource_rows`
 FROM `admin_site_resources` AS `A`
 LEFT JOIN `admin_site_organization_resource` AS `B` ON `B`.`res_id`=`A`.`id`
 LEFT JOIN `admin_site_organizations` AS `C` ON `B`.`org_id`=`C`.`id`
 LEFT JOIN `admin_site_resource_res_topic` AS `D` ON `D`.`resource_id`=`A`.`id` 
 LEFT JOIN `admin_site_resource_topics` AS `E` ON `E`.`id`=`D`.`res_topic_id`
 LEFT JOIN `admin_site_resource_audience` AS `F` ON `F`.`resource_id`=`A`.`id`
 LEFT JOIN `admin_site_audiences` AS `G` ON `G`.`id`=`F`.`audience_id`
 LEFT JOIN `admin_site_resource_curriculum_topic` AS `H` ON `H`.`resource_id`=`A`.`id`
 LEFT JOIN `admin_site_curriculum_topics` AS `I` ON `I`.`id`=`H`.`curriculum_topic_id`
 LEFT JOIN `admin_site_resource_curriculum_grade` AS `J` ON `J`.`resource_id`=`A`.`id`
 LEFT JOIN `admin_site_curriculum_grades` AS `K` ON `K`.`id`=`J`.`curriculum_grade_id`
 LEFT JOIN `admin_site_resource_curriculum_subject` AS `L` ON `L`.`resource_id`=`A`.`id`
 LEFT JOIN `admin_site_curriculum_subjects` AS `M` ON `M`.`id`=`L`.`curriculum_subject_id` 
 LEFT JOIN `admin_site_resource_res_type` AS `N` ON `N`.`resource_id`=`A`.`id`
 LEFT JOIN `admin_site_resource_types` AS `O` ON `O`.`id`=`N`.`res_type_id` 
 LEFT JOIN `admin_site_resource_res_area_location` AS `P` ON `P`.`resource_id`=`A`.`id`
 LEFT JOIN `admin_site_resource_area_locations` AS `Q` ON `Q`.`id`=`P`.`res_area_location_id` 

 WHERE `A`.`post_status`='approved' AND (
`A`.`resource_name` LIKE '%alpha%'
OR `A`.`aliases` LIKE '%alpha%' 
OR `A`.`short_desc` LIKE '%alpha%' 
OR `A`.`resource_url` LIKE '%alpha%' 
OR `A`.`other_resource_type` LIKE '%alpha%' 
OR `C`.`org_name` LIKE '%alpha%' 
OR `E`.`topic_label` LIKE '%alpha%' 
OR `G`.`audience_label` LIKE '%alpha%' 
OR `I`.`topic_label` LIKE '%alpha%' 
OR `K`.`grade_label` LIKE '%alpha%' 
OR `M`.`subject_label` LIKE '%alpha%' 
OR `O`.`type_label` LIKE '%alpha%' 
OR `Q`.`area_location_label` LIKE '%alpha%' );

Yes, I understand this query is pretty long and ugly but I guess it’s a result of all the normalization that had to be done for it. What I can’t understand is how a few rows suddenly turned to almost 68 million rows. Are the LEFT JOINS to blame for this?

Answer 1

Rewrite it as:

 SELECT COUNT(*) AS `total_num_resource_rows`
 FROM `admin_site_resources` AS `A`
 WHERE `A`.`post_status`='approved' AND
   ( `A`.`resource_name` LIKE '%alpha%'
  OR `A`.`aliases` LIKE '%alpha%' 
  OR `A`.`short_desc` LIKE '%alpha%' 
  OR `A`.`resource_url` LIKE '%alpha%' 
  OR `A`.`other_resource_type` LIKE '%alpha%' 
  OR EXISTS ( SELECT * 
              FROM `admin_site_organization_resource` AS `B` 
                JOIN `admin_site_organizations` AS `C`
                  ON `B`.`org_id`=`C`.`id`
              WHERE `B`.`res_id`=`A`.`id`
                AND `C`.`org_name` LIKE '%alpha%' 
            )
  OR EXISTS ( 
             ...
            )
  OR EXISTS ( 
             ...
            )
  ...
    )