MySQL查询缓存，复杂SQL查询

StackOverflow https://stackoverflow.com/questions/4400271

  •  25-09-2019
  •  | 
  •  

题

下面的问题让我很头疼。

我已经将 MySQL 服务器设置为使用“Query Chaching”。

set global query_cache_size = 10000000;

看来我的查询缓存有效，因为输出 

SHOW STATUS LIKE 'Qcache%';

如下

+-------------------------+----------+    
|    Variable_name        | Value    |           
+-------------------------+----------+     
| Qcache_free_blocks      | 1        |
| Qcache_free_memory      | 47223976 |
| Qcache_hits             | 6709     |                                       
| Qcache_inserts          | 2314     |                                    
| Qcache_lowmem_prunes    | 0        |                                  
| Qcache_not_cached       | 365      |                                  
| Qcache_queries_in_cache | 441      |                                   
| Qcache_total_blocks     | 960      |                                     
+-------------------------+----------+

但是，尽管如此，以下内容（复杂查询，带有内部选择等）不会被缓存。

总是需要 至少0.8秒 检索此查询的数据。我怎样才能把mysql带到 将此查询的结果存储在其缓存中?

我尝试删除内部选择，但这没有什么区别。

SELECT p.id
   AS
   project_id,
   p.code
   AS project_code,
   p.title
   AS project_title,
   p.start_date
   AS project_start_date,
   p.end_date
   AS project_end_date,
   p.modf
   AS project_modf,
   ( p.budget * (SELECT 1 / r.VALUE
                 FROM   exchange_rates r
                 WHERE  r.class_currency_id = p.class_budget_currency_id) )
   AS
   project_budget,
   (SELECT z.txt
    FROM   sys_labels z
    WHERE  z.id = ps.value_label_id
           AND z.lng = 'en')
   AS project_status,
   (SELECT z.txt
    FROM   sys_labels z
    WHERE  z.id = ps.data_label_id
           AND z.lng = 'en')
   AS project_color,
   GROUP_CONCAT(DISTINCT pt.class_topic_id)
   AS projects_thematic_area_ids,
   u.id
   AS project_owner_id
FROM   projects AS p
   LEFT JOIN projects_thematic_areas AS pt
     ON pt.project_id = p.id
   LEFT JOIN sys_users AS u
     ON u.id = p.owner_uid
   LEFT JOIN class_data s
     ON s.id = p.class_status_id
   LEFT JOIN class_data AS ps
     ON ps.id = s.id
   LEFT JOIN sys_labels AS prdz1
     ON prdz1.id = prd.value_label_id
        AND prdz1.lng = 'en'
   LEFT JOIN sys_labels AS prdz2
     ON prdz2.id = prd.data_label_id
        AND prdz2.lng = 'en'
   LEFT JOIN projects_locations AS pl
     ON pl.project_id = p.id
   LEFT JOIN class_data AS l
     ON l.id = pl.class_location_id
   LEFT JOIN class_data AS r
     ON r._lft <= l._lft
        AND r._rgt >= l._rgt
        AND r._level = 1
        AND r.class_id = 5
   LEFT JOIN class_data AS c
     ON c._lft <= l._lft
        AND c._rgt >= l._rgt
        AND c._level = 2
        AND c.class_id = 10
   LEFT JOIN projects_donors AS pd
     ON pd.project_id = p.id
   LEFT JOIN institutions AS i
     ON pd.inst_id = i.id
   LEFT JOIN class_data AS ic
     ON ic.id = i.class_country_id
   LEFT JOIN projects_deliverables AS d
     ON d.project_id = p.id
WHERE  1 = 1
   AND p.is_del = "f"
   AND p.is_active = "t"
GROUP  BY p.id
ORDER  BY p.modf DESC,
      p.code DESC

任何帮助表示赞赏......

问候

J。

有帮助吗？

解决方案

在除了以前的答案：查询缓存将不被使用，即使该查询是在那里，如果有任何选定的表的变化

但你为什么加入人这些表的时候你是不是从中选择什么？另外你可能不应该分选择任何内容，如果你能加入。

像这样将选择完全相同的：

SELECT
   p.id AS project_id,
   p.code AS project_code,
   p.title AS project_title,
   p.start_date AS project_start_date,
   p.end_date AS project_end_date,
   p.modf AS project_modf,
   p.budget * (1 / r.VALUE) AS project_budget,
   z1.txt AS project_status,
   z2.txt AS project_color,
   GROUP_CONCAT(DISTINCT pt.class_topic_id) AS projects_thematic_area_ids,
   u.id AS project_owner_id
FROM
   projects AS p
   LEFT JOIN projects_thematic_areas AS pt ON pt.project_id = p.id
   LEFT JOIN sys_users AS u ON u.id = p.owner_uid
   LEFT JOIN exchange_rates AS r ON r.class_currency_id = p.class_budget_currency_id
   LEFT JOIN class_data s ON s.id = p.class_status_id
   LEFT JOIN class_data AS ps ON ps.id = s.id
   LEFT JOIN sys_labels AS z1 ON z1.id = ps.value_label_id AND z1.lng = 'en'
   LEFT JOIN sys_labels AS z2 ON z2.id = ps.data_label_id AND z2.lng = 'en'
WHERE
   1
   AND p.is_del = "f"
   AND p.is_active = "t"
GROUP  BY
   p.id
ORDER  BY
   p.modf DESC,
   p.code DESC

当然，你已经（组合）上的所有外键索引，其中字段和组字段。考虑使用你的布尔值TINYINT或枚举场。您也可以考虑不选择该GROUP_CONCAT，所以你可能会失去在GROUP BY。也许使用INNER JOIN代替LEFT JOIN，如果你有一定的关系存在。

其他提示

您可以尝试SELECT SQL_CACHE ... FROM ...

您可以尝试一些基本的事情：

  • 阅读 查询缓存文档 确保您了解基础知识并正确设置。
  • 理想情况下，隔离您的 MySQL 数据库服务器，使其仅运行您为其提供的命令。如果您无法执行此操作，请尝试在另一台计算机上设置并运行测试。
  • 运行一个简单的查询并查看 Qcache_hits 和 Com_select 状态变量来确定查询缓存是否被命中。
  • 尝试复杂的查询并监视相同的值。如果您的查询未命中缓存，请尝试其中较小的部分，直到找出导致其未缓存的原因。如果正在缓存，则问题可能是由于查询中的任何表在查询之间更新而导致缓存的副本无效。
许可以下： CC-BY-SA 和 归因
不隶属于 StackOverflow
scroll top