How slow is FOUND_ROWS function in MySQL
From many different reasons I was always using SQL_CALC_FOUND_ROWS to get total number of records, when i was running limited query. Never especially thought about it, until i had some super performance issues with one of my projects. Obviously after quick profiling and googling i found people complaining about FOUND_ROWS() low performance. I wasn’t aware how big impact it was until i tested.
If you don’t know what FOUND_ROWS does – this is super simple example:
SELECT SQL_CALC_FOUND_ROWS * FROM tbl_name WHERE id > 100 LIMIT 10; SELECT FOUND_ROWS();
as you can tell it’s nice and simple way to check total number of results from MySQL. The other way to do that is just run it like that:
SELECT COUNT(pk) FROM tbl_name WHERE id > 100;
Which is bit more complicated if you have queries like that:
SELECT SQL_CALC_FOUND_ROWS import_date, photo, IF(ISNULL(p.formatted_address), CONCAT(street_number, " ", street_name, " ", street_type, ", ", suburb_name, " ", state, " ", postcode), formatted_address) AS formatted_address, beds, baths, cars, quote_price_low FROM `campaign` INNER JOIN `property` `p` ON `p`.`property_id` = `campaign`.`property_id` WHERE `campaign`.`import_date` >= '2015-12-25' AND `campaign`.`archive` =0 AND `campaign`.`sold` =0 AND `p`.`latitude` IS NOT NULL AND `p`.`longitude` IS NOT NULL AND (NOT EXISTS (SELECT * FROM general_dismiss WHERE general_dismiss.campaign_id = campaign.campaign_id) ) AND (NOT EXISTS (SELECT * FROM general_reject WHERE general_reject.campaign_id = campaign.campaign_id) ) AND (3959 * acos ( cos ( radians(-27.5649) ) * cos( radians( p.latitude ) ) * cos( radians( p.longitude ) - radians(152.968) ) + sin ( radians(-27.5649) ) * sin( radians( p.latitude ) ) ) ) <=5 ORDER BY `import_date` DESC LIMIT 10
But… what’s the actual time difference between this 2 queries – in database having around 640k records in MySQL. First query –
FOUND_ROWS was taking around 7.3932 seconds to run, second one 0.5466.
I wrote quick function to help replace this tricky part in MySQL query