我有一個相當大的 SQL 查詢被 GROUP BY 和 ORDER BY 語句減慢。 優化的最佳方法是什么？

Question

優化此查詢的最佳方法是什么，因為它最多需要 12 秒才能執行？

為查詢的大小道歉，它幾乎立即執行，沒有 group by 和 order by 語句。

我對 SQL 優化還很陌生，我只編碼了大約一年。 我沒有寫這個查詢，但我需要修復它。

SELECT
`products`.*,
`product_alternative`.`alternative_product_code`,
`product_alternative`.`id` as `product_alternative_id`,
`vat_rate`.`rate` AS `vat_rate`,
`product_images`.`filename` AS `product_image_file`,
`product_docs`.`filename` AS `product_doc_file`,
`suppliers`.`supplier_code`,
`suppliers`.`name` AS `supplier_name`,
`commission`.`commission` AS `supplier_commission`,
`categories`.`name` AS `category_name`,
`sub_categories`.`name` AS `subcategory_name`,
IF(`products`.`product_doc_id` = 0,
 NULL,
 CONCAT(
"/product-docs/",
`products`.`id`
)) AS `product_doc_url`,IF(`products`.`product_image_id` = 0, NULL, CONCAT(
"/product-images/",
`products`.`id`,
"/original/"
)) AS `original_image_url`,IF(`products`.`product_image_id` = 0, NULL, CONCAT(
"/product-images/",
`products`.`id`,
"/medium/"
)) AS `medium_image_url`,IF(`products`.`product_image_id` = 0, NULL, CONCAT(
"/product-images/",
`products`.`id`,
"/thumb/"
)) AS `thumb_image_url`,CASE
WHEN
    `group_favourite`.`id` IS NOT NULL
    THEN 1 ELSE 0
END
AS `group_favourite`
,CASE
WHEN
    `product_favourite`.`org_id` != 0
    AND `product_favourite`.`org_id` IS NOT NULL
    THEN 1 ELSE 0
END
AS `favourite`
,CASE
WHEN
    `product_favourite`.`org_id` != 0
    AND `product_favourite`.`org_id` IS NOT NULL
    THEN 1 ELSE 0
END
AS `favourite`,
`product_favourite`.`needs_authorisation`,

 IF(`customer_personal_favourite`.`id` IS NOT NULL,1,0) AS `is_personal_favourite`
 FROM
`products` 

LEFT JOIN `vat_rate` ON `products`.`vat_rate_id` = `vat_rate`.`id`
LEFT JOIN `product_images` ON `products`.`product_image_id` = `product_images`.`id`
LEFT JOIN `product_docs` ON `products`.`product_doc_id` = `product_docs`.`id`

LEFT JOIN `orgs` AS `suppliers` ON `products`.`supplier_id` = `suppliers`.`id`

LEFT JOIN `commission` ON
    `suppliers`.`id` = `commission`.`org_id`
    AND `commission`.`status` = 1
    AND DATE(`commission`.`start_date`) <= DATE('2020-07-15')
    AND (
        DATE(`commission`.`end_date`) >= DATE('2020-07-15')
        OR `commission`.`end_date` = '0000-00-00'
    )

LEFT JOIN `categories` AS `categories` ON
    `products`.`category_id` = `categories`.`id`


LEFT JOIN `categories` AS `sub_categories` ON
    `products`.`sub_category_id` = `sub_categories`.`id`

LEFT JOIN `product_attribute_product` ON
    `product_attribute_product`.`product_id` = `products`.`id`
LEFT JOIN `product_alternative` ON
    `product_alternative`.`product_original_id` = `products`.`original_id`
JOIN `orgs` AS `customer` ON
`customer`.`id` IN (320)

LEFT JOIN `product_supplier_account` ON
`product_supplier_account`.`product_id` = `products`.`id`

JOIN `supplier_account` AS `default_supplier_account` ON
`default_supplier_account`.`supplier_id` = `products`.`supplier_id`
AND `default_supplier_account`.`is_default` = 1
AND `default_supplier_account`.`is_deleted` = 0

JOIN `customer_to_supplier` ON
`customer_to_supplier`.`supplier_id` = `products`.`supplier_id`
AND `customer_to_supplier`.`customer_id` IN (`customer`.`id`)
AND `customer_to_supplier`.`allow_access` = 1
AND `customer_to_supplier`.`status` = 1

JOIN `customer_to_supplier_account` ON
    `customer_to_supplier_account`.`customer_to_supplier_id` = `customer_to_supplier`.`id`
    AND (
        `customer_to_supplier_account`.`supplier_account_id` =                                    `product_supplier_account`.`supplier_account_id`
        OR `customer_to_supplier_account`.`supplier_account_id` = `default_supplier_account`.`id`
    )
    AND `customer_to_supplier_account`.`allow_access` = 1
    AND `customer_to_supplier_account`.`is_deleted` = 0
LEFT JOIN `statement_products_categories` ON
`statement_products_categories`.`product_id` = `products`.`original_id`
LEFT JOIN `product_favourite` ON
    `product_favourite`.`product_id` = `products`.`original_id`
    AND `product_favourite`.`org_id` IN (`customer`.`id`)
    LEFT JOIN `group_favourite` AS `group_favourite` ON
`products`.`original_id` = `group_favourite`.`product_id`
AND `group_favourite`.`group_id` IN (37)
LEFT JOIN `budget` AS `category_budget` ON
    `category_budget`.`org_id` IN (`customer`.`id`)
    AND `category_budget`.`budget_type_id` = 1
    AND `category_budget`.`category_id` = `products`.`category_id`
    AND
        `category_budget`.`start_date` <= CURDATE()
        AND (
            `category_budget`.`final_date` >= CURDATE()
            OR `category_budget`.`final_date` IS NULL
        )
LEFT JOIN `budget` AS `supplier_budget` ON
    `supplier_budget`.`org_id` IN (`customer`.`id`)
    AND `supplier_budget`.`budget_type_id` = 3
    AND `supplier_budget`.`supplier_id` = `products`.`supplier_id`
    AND
        `supplier_budget`.`start_date` <= CURDATE()
        AND (
            `supplier_budget`.`final_date` >= CURDATE()
            OR `supplier_budget`.`final_date` IS NULL
        )
LEFT JOIN `budget` AS `product_budget` ON
    `product_budget`.`org_id` IN (`customer`.`id`)
    AND `product_budget`.`budget_type_id` = 2
    AND
        `product_budget`.`start_date` <= CURDATE()
        AND (
            `product_budget`.`final_date` >= CURDATE()
            OR `product_budget`.`final_date` IS NULL
        )
LEFT  JOIN `customer_personal_favourite` ON
`customer_personal_favourite`.`org_id` = `customer`.`id`
AND `customer_personal_favourite`.`product_original_id` = `products`.`original_id`

WHERE
0 OR (1  AND `products`.`status` = 1  AND ((
    `products`.`parent_id` = 0
    OR `products`.`parent_id` IS NULL
)   OR (
    `product_favourite`.`org_id` IS NOT NULL
    AND `product_favourite`.`org_id` != 0
)  )  AND `products`.`id` = `products`.`current_id` )
GROUP BY `products`.`id`
ORDER BY
`favourite` DESC,

`active_date` ASC,
    `products`.`code` ASC,
    `products`.`name` ASC,
    `products`.`parent_id` ASC

 LIMIT 0, 20 

Answer 1

我曾經遇到過這個問題，當時我決定寫一篇關於這個的文章，如果有人想看看我是如何實現的，我已經從不同的來源編譯並優化了我的查詢。 請看看這個。 https://junaidtechblog.wordpress.com/2019/09/04/optimize-sql-query-groupby-having/

Answer 2

      AND  DATE(`commission`.`start_date`) <= DATE('2020-07-15')
      AND  ( DATE(`commission`.`end_date`) >= DATE('2020-07-15')
              OR  `commission`.`end_date` = '0000-00-00' 

有幾個問題：

• OR很難優化。 找到一些方法來避免它——例如刪除這些行，或者使用一個遙遠的未來而不是過去的結束日期。
• 不要隱藏 function 調用中的列。
• 讓我們看看SHOW CREATE TABLE 。
• 您不需要DATE()圍繞文字“2020-07-15”。

另一方面，這沒問題： product_budget 。 開始日期 <= start_date ()

ORDER BY `favourite` DESC,
         `active_date` ASC,

除非您有 MySQL 8.0，否則沒有索引可以處理混合了 DESC 和 ASC 的ORDER BY 。 （由於其他原因，可能無法使用索引。）

WHERE 0 OR (1  AND ...

動態構建查詢，而不是使用像這樣的組合。

LEFT JOIN ... ON ... foo = 1

ON應該 state 表是如何相關的。 foo = 1感覺更像是應該在WHERE子句中的東西（用於過濾）。 在ON和WHERE之間移動它可能會改變生成的行集。 但我質疑這是否應該是JOIN （而不是LEFT JOIN ）或者測試應該在WHERE中，或者它是否真的像你寫的那樣有效。

請提供EXPLAIN SELECT...具有最大“行”的行可能是要關注的表。

GROUP BY  `products`.`id`

很可能導致損壞的 output。 閱讀“only_full_group_by”。

這會導致潛在的加速。

1. 做最少的工作來獲得一個滿足ORDER BY和LIMIT的id列表。

2. 然后JOIN所有其他表，包括返回products ，以獲取其他列。 這可能會加快速度並擺脫上述錯誤。

3. 在哪里實際將LEFT JOIN更改為子查詢：

    SELECT... `vat_rate`.`rate` AS `vat_rate`, ... LEFT JOIN `vat_rate` ON `products`.`vat_rate_id` = `vat_rate`.`id`...

-->

    SELECT ...
        ( SELECT `rate` FROM vat_rate WHERE id = `products`.`vat_rate_id`
        )  AS `vat_rate`,
        ...
    ...

完成大部分工作，然后回來尋求更多建議。