Question

SELECT tmp.question_set_id,
       qs.sort_order, qa.type, qa.entities, qa.segment, qa.number_entities, qa.display,
       q.content, qa.value, m.title as media_title, m.content as media_content 
FROM question_set_answer qsa 
LEFT JOIN 
       (SELECT * FROM question_set_answer ORDER BY RAND()) tmp ON (qsa.id= tmp.id) 
JOIN question_set qs ON qs.id = tmp.question_set_id 
JOIN question_answer qa ON qa.id = tmp.question_answer_id 
JOIN question q ON q.id = qa.question_id 
LEFT JOIN 
       (SELECT component_id, title, content FROM media WHERE component ='question_answer') 
        m ON m.component_id = tmp.question_answer_id 
        WHERE qs.component_id = 142 
        AND qs.component = 'game_level' 
GROUP BY tmp.question_set_id 
ORDER BY qsa.id

Answer 1

(SELECT * FROM question_set_answer ORDER BY RAND())

AAAAAAAAAAAAAAAAAAAAAA！

的 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA！ 的

* ahem *对不起，我在那里失去了镇静。

ORDER BY RAND()表示必须扫描和排序整个表。即使您的查询的其余部分正确地匹配每个索引，如果表具有任何大量数据，这将是您的瓶颈。从您提出这个问题的事实来看，我认为您遇到了性能问题。

除了这一般建议之外，您能否为该查询发布EXPLAIN的输出？

查询需要优化，有人可以帮忙吗？

1 个答案: