如何在不查询的情况下获得更好的性能

Question

here is my custom sql query ;

SELECT id,
       post_title,
       post_content,
       comment_count,
       post_date,
       post_status,
       post_name
FROM   wp_posts
WHERE  post_type = 'post'
       AND post_status = 'publish'
       AND id NOT IN (SELECT DISTINCT post_id
                      FROM   wp_postmeta
                      WHERE  meta_key = 'visible_headlane'
                             AND meta_value = 'On')
       AND id NOT IN (SELECT DISTINCT post_id
                      FROM   wp_postmeta
                      WHERE  meta_key = 'visible_homepage'
                             AND meta_value = 'On')
ORDER  BY post_date DESC
LIMIT  11, 32  

i can not use, LIMIT in NOT IN Query so, there is any way to use, or limit SELECT DISTINCT sql query, or any idea ?

Note : I need to optimise this query because, it takes so long , like a slow query.

Thanks.

Answer 1

Switch to NOT EXISTS( SELECT 1 FROM ... ) -- this variation does not need to compute the entire list of things; it only needs to check for the absence. LEFT JOIN .. ON .. IS NULL is also more efficient.

Furthermore, the standard schema for wp_postmeta is quite inefficient; see my suggestions .

Meanwhile, keep it as two tests, unlike the suggestion by 'holder'.

The EXISTS would look something like

AND NOT EXISTS ( SELECT 1 FROM wp_postmeta
                    WHERE post_id = wp_posts.id
                      AND meta_key = 'visible_headlane'
                      AND meta_value = 'On' )

which would make good use of the PRIMARY KEY(post_id, meta_key) that I recommend.

(Is it really 'headlane', not 'headline'?)

Answer 2

Maybe something like this as jarlh suggested

SELECT 
ID,post_title,post_content,comment_count,post_date,post_status,post_name
FROM wp_posts t1
left join  (
SELECT DISTINCT post_id from wp_postmeta
WHERE meta_key IN ('visible_headlane','visible_homepage') AND meta_value = 'On' ) t2 on t1.ID = t2.post_id

WHERE post_type = 'post'
AND post_status = 'publish'
AND t2.post_id is null
ORDER BY post_date DESC
LIMIT 11, 32