使用帶IN子句的嵌套子查詢提高MySQL查詢的性能

Question

我想提高以下查詢的性能。

SELECT DISTINCT StringId 
FROM translations 
WHERE status = 1 
AND  TranslationId IN 
( 
    SELECT Max(TranslationId) 
    FROM translations 
    WHERE languageId = 2 
    AND  TranslationId > 0 
    GROUP BY StringId; 

此查詢當前需要永久因爲In子句。

下面的子查詢

SELECT Max(TranslationId) 
FROM translations 
WHERE languageId = 2 
AND  TranslationId > 0 
GROUP BY StringId; 

收益16,000行，並與16,000行子句什麼是這麼長時間。

，我想出了一個改進的查詢是

SELECT DISTINCT t1.StringId 
FROM translations t1 
INNER JOIN 
( 
    SELECT Max(TranslationId) MaxTranslationId, 
      StringId 
    FROM translations 
    WHERE languageId = 2 
    AND  TranslationId > 0 
    GROUP BY StringId 
) t2 ON t1.StringId = t2.StringId 
     AND t1.TranslationId = t2.MaxTranslationId 
WHERE status = 1; 

我希望內部聯接將返回只喜歡只用在子句同時匹配的StringID和Max（TranslationId）的行。

如果這是一個正確的方法來處理，有人可以給我一個答案嗎？

Answer 1

以下是使用NOT EXISTS另一種方法：

SELECT DISTINCT t.StringId 
FROM translations t 
WHERE status = 1 AND 
     NOT EXISTS (select 1 
        from translations t2 
        where t2.languageId = 2 and 
         t2.StringId = t.StringId and 
         t2.translationId > t.translationId 
       ) AND 
     EXISTS (select 1 
       from translations t2 
       where t2.languageId = 2 and 
        t2.StringId = t.StringId and 
        t2.translationId > 0 
       ) 

出於性能考慮，你想對translations(StringId, languageId, translationId)的索引。這消除了其中一個聚合，將其替換爲索引查找，這可能會更快。

編輯：上面應該工作。但是，這也可能是相對高效的：

SELECT DISTINCT StringId 
FROM translations t 
WHERE status = 1 AND 
     t.TranslationId = (SELECT Max(t2.TranslationId) 
          FROM translations t2 
          WHERE t2.languageId = 2 AND t2.TranslationId > 0 AND 
            t2.StringId = t.StringId 
         );