Mysql 我有一个地狱般的SQL查询I'；我想优化一下。让'；让我们谈谈吧

所以，这里有一个问题可能会让SQL专家们跳到我身上说我懒惰，但我被难倒了。我们的在线商店今早崩溃并烧毁，这是可疑的查询。我整天都在考虑这个问题，还没有想出任何天才的优化方法。我能得到一些帮助吗？有什么重要的指标吗？如何重组？我意识到这就像问墙的另一边是什么，然后给你一个指向另一个方向的望远镜，但我觉得值得一试：

SELECT DISTINCT (SELECT filename FROM (SELECT DISTINCT y.value AS label, x.value AS filename 
                        FROM   `catalog_product_super_link` AS z 
                        INNER JOIN `catalog_product_entity_varchar` AS y 
                        ON z.product_id = y.entity_id 
                        INNER JOIN `catalog_product_entity_varchar` AS x 
                        ON z.product_id = x.entity_id 
                        WHERE  parent_id = (SELECT entity_id 
                                             FROM   `catalog_product_entity`
                                             WHERE  sku LIKE 'F11-ARC-7710%' 
                                             LIMIT  0, 1) 
                        AND y.attribute_id = (SELECT attribute_id 
                                               FROM   `eav_attribute` 
                                               WHERE  attribute_code = 'image_label' 
                                               AND entity_type_id = (SELECT entity_type_id FROM `eav_entity_type` WHERE entity_type_code = 'catalog_product') LIMIT 0, 1) 
                        AND x.attribute_id = (SELECT attribute_id 
                                               FROM   `eav_attribute` 
                                               WHERE  attribute_code = 'image' 
                                               AND entity_type_id = (SELECT entity_type_id FROM `eav_entity_type` WHERE entity_type_code = 'catalog_product'))) AS images WHERE c.value LIKE CONCAT(label,'%') LIMIT 0, 1) AS image,
                        pricing_value, 
                        is_percent, 
                        value_index, 
                        c.value AS label,
                        d.sort_order AS sort_order
                        FROM   `catalog_product_super_attribute_pricing` AS a 
                        INNER JOIN `catalog_product_super_attribute_label` AS b 
                        ON a.product_super_attribute_id = b.product_super_attribute_id 
                        INNER JOIN `eav_attribute_option_value` AS c 
                        ON value_index = c.option_id 
                        INNER JOIN `eav_attribute_option` AS d
                        ON c.option_id = d.option_id 
                        WHERE  a.product_super_attribute_id = (SELECT product_super_attribute_id 
                                                               FROM   `catalog_product_super_attribute` 
                                                               WHERE  product_id = 5928 
                                                               AND attribute_id = 143 LIMIT 0, 1) 

                        UNION ALL

                        SELECT DISTINCT (SELECT filename FROM (SELECT DISTINCT y.value AS label, x.value AS filename 
                                         FROM   `catalog_product_super_link` AS z 
                                         INNER JOIN `catalog_product_entity_varchar` AS y 
                                         ON z.product_id = y.entity_id 
                                         INNER JOIN `catalog_product_entity_varchar` AS x 
                                         ON z.product_id = x.entity_id 
                                         WHERE  parent_id = (SELECT entity_id 
                                                             FROM   `catalog_product_entity`
                                                             WHERE  sku LIKE 'F11-ARC-7710%' 
                                                             LIMIT  0, 1) 
                                         AND y.attribute_id = (SELECT attribute_id 
                                                               FROM   `eav_attribute`
                                                               WHERE  attribute_code = 'image_label' 
                                                               AND entity_type_id = (SELECT entity_type_id FROM `eav_entity_type` WHERE entity_type_code = 'catalog_product') LIMIT 0, 1) 
                                         AND x.attribute_id = (SELECT attribute_id 
                                                               FROM   `eav_attribute` 
                                                               WHERE  attribute_code = 'image' 
                                                               AND entity_type_id = (SELECT entity_type_id FROM `eav_entity_type` WHERE entity_type_code = 'catalog_product'))) AS images
                                                               WHERE label LIKE CONCAT((SELECT value FROM `eav_attribute_option_value` WHERE option_id = c.value LIMIT 0, 1),'%') LIMIT 0, 1) AS image,
                                        0 AS pricing_value, 
                                        0 AS is_percent, 
                                        c.value AS value_index, 
                                        (SELECT value FROM `eav_attribute_option_value` WHERE  option_id = c.value LIMIT  0, 1) AS label,
                                        (SELECT sort_order FROM `eav_attribute_option` WHERE option_id = c.value LIMIT 0, 1) AS sort_order
                        FROM   `catalog_product_entity` AS a 
                        INNER JOIN `cataloginventory_stock_status` AS b 
                        ON a.entity_id = b.product_id 
                        INNER JOIN `catalog_product_entity_int` AS c 
                        ON a.entity_id = c.entity_id 
                        INNER JOIN `cataloginventory_stock_item` AS d 
                        ON a.entity_id = d.product_id 
                        WHERE  c.attribute_id = (SELECT attribute_id 
                                                 FROM   `eav_attribute` 
                                                 WHERE  attribute_code = 'choose_size' 
                                                 AND entity_type_id = (SELECT entity_type_id FROM `eav_entity_type` WHERE entity_type_code = 'catalog_product') LIMIT 0, 1) 
                        AND a.entity_id IN (SELECT DISTINCT product_id 
                                            FROM   `catalog_product_super_link` 
                                            WHERE  parent_id = (SELECT entity_id FROM `catalog_product_entity` WHERE sku LIKE 'F11-ARC-7710%' LIMIT  0, 1)) 
                        AND (b.qty > 0 OR d.manage_stock = 0) 
                        AND (SELECT value 
                             FROM   `eav_attribute_option_value` 
                             WHERE  option_id = c.value 
                             LIMIT  0, 1) NOT IN (SELECT c.value 
                                                  FROM `catalog_product_super_attribute_pricing` AS a
                                                  INNER JOIN `catalog_product_super_attribute_label` AS b ON a.product_super_attribute_id = b.product_super_attribute_id 
                                                  INNER JOIN `eav_attribute_option_value` AS c ON value_index = c.option_id 
                                                  WHERE  a.product_super_attribute_id = (SELECT product_super_attribute_id FROM `catalog_product_super_attribute`
                                                                                 WHERE product_id = 5928 
                                                                                         AND attribute_id = 143))
                        ORDER BY sort_order

提前谢谢

噢，天啊

除了WHERE子句中的每一列上的索引之外，我没有什么可以提供的，我还解释了查看是否有表扫描的计划

我的经验法则是，七个或更多的连接将表现不佳。我计算的内部连接子句比你的多。我对这个问题会有严重的保留。有可能去规范化吗？风景怎么样？这对你有帮助吗？你比我更了解你的模式和问题。

快速回答

你能用PLSQL的等效语言来做吗？ 你能消除这些子查询吗

首先执行这些操作并存储结果。 假设它们返回一行

---

然后，编写一个较小的查询，传入变量。

前几天我回答的这个问题可能会对您有所帮助：

---

它是关于使嵌套查询更高效。我怀疑，如果您单独处理嵌套选择，然后使用结果形成一个更简单的查询，这会有所帮助。允许的巡演查询要复杂得多，所以可能会有更多，但希望这只是一个开始。

如果没有更多的信息，甚至没有您的测试台，这几乎是不可能的

但除了按照其他人的建议运行解释计划和添加索引之外，我还看到一些可能可以从中提取出来以改进的东西：

例如，您重复了子选择：例如：

(SELECT entity_id 
FROM   `catalog_product_entity`
WHERE  sku LIKE 'F11-ARC-7710%' 
LIMIT  0, 1) 

考虑将它们拉到FROM子句的最高级别，然后加入以获得结果。一次不多。如果对3或4个变体执行此操作，您应该会看到改进

---

另一件看起来不太正确的事情是，在子选择中嵌套的非常深。。。您可能不需要这些，它们可能会增加开销。

此查询太长、太复杂，无法花时间理解，因此可以对其进行优化。您能给出输入数据、表结构和所需输出的（小！）示例吗？PLSQL是Oracle特有的。标签上写着MySQL。但是存储过程是个好主意，这不是个坏主意。所有这些子查询都是使这个特定查询在各种开发/登台/实时环境中始终工作所必需的（我们在Magento上，任何在Magento上的人都会理解——IDs的变化）。但是分解它们只会将其转换为7个查询而不是一个…在这种情况下，我相信7个查询将比一个嵌套查询更快。还应该能够检测您所在的环境并有条件地构建正确的查询。通过拆分查询并缓存子查询的结果来解决，这样子查询只执行一次。