mysql优化

我有一个包含400多个事件的“事件”表，其中包含以下关系：一对多的事件（45000行），一对多的事件（1000行）

如果有人建议重新改进这个冗长而缓慢的查询（我的sql技能相当差），我将不胜感激：

编辑： 这是一个 事件搜索查询，需要同时按类别表和日期表进行搜索。下表结构：

事件日期：

id开始日期结束日期开始时间结束时间事件id

---

事件类别：

id event\u id category\u id

一方面，摆脱像“%London%”这样的

和events.city

，让你的位置定位依赖于县id（我希望是外键）

，摆脱像“%London%”这样的

和events.city

，使您的定位目标依赖于县id（我希望是外键）

我将您的查询重新编写为：

  SELECT e.id, 
         e.name AS event, 
         e.listing_type AS TYPE , 
         ec.category_id,                 
         e.location,
         e.summary, 
         co.name AS county, 
         e.postcode, 
         e.picture, 
         e.venue, 
         e.lat, 
         e.lng 
    FROM EVENTS e
    JOIN EVENT_DATES ed ON ed.event_id = e.id
    JOIN (SELECT t.event_id,
                 GROUP_CONCAT( DISTINCT t.category_id ) AS category_id
            FROM EVENT_CATEGORIES t
        GROUP BY t.event_id) ec ON ec.event_id = e.id
    JOIN COUNTIES co ON c.id = e.county_id 
   WHERE e.preview = 0
     AND ed.start_date > DATE( NOW( ) )
     AND e.city LIKE '%London%'
ORDER BY e.id DESC

你有：

• 连接到CATEGORIES表，但未使用它
• 虽然OP中的GROUP BY在MySQL上有效，但此版本是SQL标准
• 如果

  EVENTS.city

  （即：'%London%'）上有索引，则通过左侧带有通配符的LIKE搜索将无法使用索引

建议/建议

• 为连接条件列编制索引（如果尚未编制），然后查看WHERE子句中的列的索引

---

我将您的查询重新编写为：

  SELECT e.id, 
         e.name AS event, 
         e.listing_type AS TYPE , 
         ec.category_id,                 
         e.location,
         e.summary, 
         co.name AS county, 
         e.postcode, 
         e.picture, 
         e.venue, 
         e.lat, 
         e.lng 
    FROM EVENTS e
    JOIN EVENT_DATES ed ON ed.event_id = e.id
    JOIN (SELECT t.event_id,
                 GROUP_CONCAT( DISTINCT t.category_id ) AS category_id
            FROM EVENT_CATEGORIES t
        GROUP BY t.event_id) ec ON ec.event_id = e.id
    JOIN COUNTIES co ON c.id = e.county_id 
   WHERE e.preview = 0
     AND ed.start_date > DATE( NOW( ) )
     AND e.city LIKE '%London%'
ORDER BY e.id DESC

你有：

• 连接到CATEGORIES表，但未使用它
• 虽然OP中的GROUP BY在MySQL上有效，但此版本是SQL标准
• 如果

  EVENTS.city

  （即：'%London%'）上有索引，则通过左侧带有通配符的LIKE搜索将无法使用索引

建议/建议

• 为连接条件列编制索引（如果尚未编制），然后查看WHERE子句中的列的索引

---

我认为您应该在更小的范围内（作为FROM子句中的子查询）执行GROUP\u CONCAT操作，这样可以很好地提高整体性能。因此，也将从查询中删除表

categories

，因为它只会造成混乱（除非您的表处于语义不一致的状态，以至于您的事件\ u类别中的类别id值不会出现在定义类别表中）

这些观察结果导致：

SELECT e.id AS id,
       e.name AS event,
       e.listing_type AS type,
       cat.category_id AS category_id,
       e.location AS location,
       e.summary AS summary,
       counties.name AS county,
       e.postcode AS postcode,
       e.picture AS picture,
       e.venue AS venue,
       e.lat AS lat,
       e.lng AS lng
  FROM Events AS e
  JOIN event_dates AS d ON e.id = d.id AND d.start_date > DATE(NOW())
  JOIN (SELECT id, GROUP_CONCAT(category_id) AS category_id
          FROM event_categories
         GROUP BY id
       ) AS cat ON e.id = cat.id
  JOIN counties ON e.county_id = counties.id
 WHERE e.preview = 0
   AND e.city LIKE '%London%'
 ORDER BY events.id DESC

如果查询中确实需要categories表，则它属于子查询：

SELECT e.id AS id,
       e.name AS event,
       e.listing_type AS type,
       cat.category_id AS category_id,
       e.location AS location,
       e.summary AS summary,
       counties.name AS county,
       e.postcode AS postcode,
       e.picture AS picture,
       e.venue AS venue,
       e.lat AS lat,
       e.lng AS lng
  FROM Events AS e
  JOIN event_dates AS d ON e.id = d.id AND d.start_date > DATE(NOW())
  JOIN (SELECT id, GROUP_CONCAT(category_id) AS category_id
          FROM event_categories AS ec
          JOIN categories AS ct ON ec.category_id = ct.category_id
         GROUP BY id
       ) AS cat ON e.id = cat.id
  JOIN counties ON e.county_id = counties.id
 WHERE e.preview = 0
   AND e.city LIKE '%London%'
 ORDER BY events.id DESC

---

我认为您应该在更小的范围内（作为FROM子句中的子查询）执行GROUP_CONCAT操作，这可能会提高整体性能。因此，也将从查询中删除表

categories

，因为它只会造成混乱（除非您的表处于语义不一致的状态，以至于您的事件\ u类别中的类别id值不会出现在定义类别表中）

这些观察结果导致：

SELECT e.id AS id,
       e.name AS event,
       e.listing_type AS type,
       cat.category_id AS category_id,
       e.location AS location,
       e.summary AS summary,
       counties.name AS county,
       e.postcode AS postcode,
       e.picture AS picture,
       e.venue AS venue,
       e.lat AS lat,
       e.lng AS lng
  FROM Events AS e
  JOIN event_dates AS d ON e.id = d.id AND d.start_date > DATE(NOW())
  JOIN (SELECT id, GROUP_CONCAT(category_id) AS category_id
          FROM event_categories
         GROUP BY id
       ) AS cat ON e.id = cat.id
  JOIN counties ON e.county_id = counties.id
 WHERE e.preview = 0
   AND e.city LIKE '%London%'
 ORDER BY events.id DESC

如果查询中确实需要categories表，则它属于子查询：

SELECT e.id AS id,
       e.name AS event,
       e.listing_type AS type,
       cat.category_id AS category_id,
       e.location AS location,
       e.summary AS summary,
       counties.name AS county,
       e.postcode AS postcode,
       e.picture AS picture,
       e.venue AS venue,
       e.lat AS lat,
       e.lng AS lng
  FROM Events AS e
  JOIN event_dates AS d ON e.id = d.id AND d.start_date > DATE(NOW())
  JOIN (SELECT id, GROUP_CONCAT(category_id) AS category_id
          FROM event_categories AS ec
          JOIN categories AS ct ON ec.category_id = ct.category_id
         GROUP BY id
       ) AS cat ON e.id = cat.id
  JOIN counties ON e.county_id = counties.id
 WHERE e.preview = 0
   AND e.city LIKE '%London%'
 ORDER BY events.id DESC

---

post explain plan、show INDEX from和describe plus简要说明您正在尝试完成的工作-谢谢：）使用显式连接表示法连接表。我支持f00-我怀疑连接的排序可能效率低下，至少是这样。此外，正如coolgeek指出的，“LIKE”（和其他字符串比较）往往相对较慢，并且会影响性能。第三个考虑因素是组中的DISTINCT-如果事件类别是一个真正的多对多表，那么单个事件条目不应该有重复的类别条目，并且在该表上使用多值PK将允许您在此查询中删除DISTINCT。post explain plan，show index from和descripe plus简要描述您正在尝试完成的任务-谢谢：）使用显式连接表示法连接表。我支持f00-我怀疑连接的排序可能效率低下，至少是这样。此外，正如coolgeek指出的，“LIKE”（和其他字符串比较）往往相对较慢，并且会影响性能。第三个考虑因素是组中的DISTINCT-如果事件类别是一个真正的多对多表，那么单个事件条目不应该有重复的类别条目，并且在该表上使用多值PK将允许您在该查询中删除DISTINCT。+1：关于如何包括类别的说明。。。和我的格式非常相似。+1：关于如何包含类别的注释。。。格式和我的非常相似。谢谢你们。出于某种奇怪的原因，我的查询时间从6秒缩短到了58秒。但真正起作用的是，正如您所说的那样，对连接进行适当的索引（现在是0.05秒）。干杯，谢谢你们。出于某种奇怪的原因，我的查询时间从6秒缩短到了58秒。但真正起作用的是，正如您所说的那样，对连接进行适当的索引（现在是0.05秒）。干杯