MySQL 左外连接查询太慢

Question

我有 3 张桌子：

订单（5429850 行），order_lines（PK id，FK order_id，18530647 行），order_line_allocations（PK id，FK order_line_id，112594 行）

order_lines和order_line_allocations存储帐户相关信息。 对于单账户情况，我们将账户数据存储在 order_lines 表中，对于多账户用例，我们将数据存储在分配表中。

我正在尝试编写一个查询来获取满足用户帐户限制的订单。

SELECT `orders`.`id` FROM `orders` 
LEFT JOIN `order_lines` ON `order_lines`.`order_id` = `orders`.`id` 
LEFT JOIN order_line_allocations ON order_line_allocations.order_line_id = order_lines.id
LEFT JOIN (select sec_accounts.id from accounts sec_accounts
  WHERE (sec_accounts.account_type_id = 344 AND ((`sec_accounts`.`segment_1` = 'MS')))
) as non_alloc_accounts ON non_alloc_accounts.id = order_lines.account_id AND order_lines.allocation_count = 0
LEFT JOIN (select sec_accounts.id from accounts sec_accounts
  WHERE (sec_accounts.account_type_id = 344 AND ((`sec_accounts`.`segment_1` = 'MS')))
) as alloc_accounts ON alloc_accounts.id = order_line_allocations.account_id 
WHERE (`orders`.line_count = 0 OR
       `orders`.account_type_id IS NULL OR
       `orders`.account_type_id IN (NULL) OR 
       (non_alloc_accounts.id is not null) OR
       (alloc_accounts.id is not null)
) ORDER BY `orders`.`id` ASC LIMIT 90 \G;

此查询需要 7 秒才能完成，太长了

但是，对于其他用户：

SELECT `orders`.`id` FROM `orders` 
LEFT JOIN `order_lines` ON `order_lines`.`order_id` = `orders`.`id` 
LEFT JOIN order_line_allocations ON order_line_allocations.order_line_id = order_lines.id
LEFT JOIN (select sec_accounts.id from accounts sec_accounts
  WHERE (sec_accounts.account_type_id = 8 AND ((`sec_accounts`.`segment_1` = '2D')))
) as non_alloc_accounts ON non_alloc_accounts.id = order_lines.account_id AND order_lines.allocation_count = 0
LEFT JOIN (select sec_accounts.id from accounts sec_accounts
  WHERE (sec_accounts.account_type_id = 8 AND ((`sec_accounts`.`segment_1` = '2D')))
) as alloc_accounts ON alloc_accounts.id = order_line_allocations.account_id 
WHERE (`orders`.line_count = 0 OR
       `orders`.account_type_id IS NULL OR
       `orders`.account_type_id IN (NULL) OR 
       (non_alloc_accounts.id is not null) OR
       (alloc_accounts.id is not null)
) ORDER BY `orders`.`id` ASC LIMIT 90 \G;

^ 这只需 0.03 秒。 请注意，不同之处在于 LEFT JOINS 的 where 子句。 特别是“=8”与“=344”在两个地方。

匹配 8 和 '2D' 的帐户数量是 175667，对于 344 和 'MS' 它只是 5648 。

我在所有表上都有所需的索引

表索引：

订单

PRIMARY KEY (`id`),
UNIQUE KEY `index_orders_on_po_number` (`po_number`),
KEY `index_orders_on_account_type_id` (`account_type_id`),
KEY `index_orders_on_created_by` (`created_by_id`),
KEY `index_orders_on_last_exported_at` (`last_exported_at`),
KEY `index_orders_on_status` (`status`),
KEY `index_orders_on_updated_at` (`updated_at`),
KEY `index_orders_on_updated_by` (`updated_by_id`),
KEY `index_orders_on_created_at_and_status` (`created_at`,`status`),
KEY `index_status_supplier_id` (`status`,`supplier_id`),
KEY `index_orders_on_line_count` (`line_count`)

order_lines

PRIMARY KEY (`id`),
UNIQUE KEY `index_order_lines_on_bulk_price_id` (`bulk_price_id`),
KEY `index_order_lines_on_account_type_id` (`account_type_id`),
KEY `index_order_lines_on_created_at` (`created_at`),
KEY `index_order_lines_on_created_by` (`created_by_id`),
KEY `index_order_lines_on_order_id` (`order_id`),
KEY `index_order_lines_on_status` (`status`),
KEY `index_order_lines_on_updated_at` (`updated_at`),
KEY `index_order_lines_on_updated_by` (`updated_by_id`),
KEY `index_order_lines_on_order_id_and_position` (`order_id`,`position`),
KEY `index_order_lines_on_order_id_and_line_num` (`order_id`,`line_num`),
KEY `index_ol_on_header_id_reporting_total_savings_pct_created_at` (`order_id`,`reporting_total`,`savings_pct`,`created_at`),
KEY `index_ol_on_order_id_supplier_id_reporting_total` (`order_id`,`supplier_id`,`reporting_total`),
KEY `index_ol_on_order_id_commodity_id_reporting_total_ela_id` (`order_id`,`commodity_id`,`reporting_total`,`extra_line_attribute_id`),
KEY `index_order_lines_on_allocation_count_and_account_id` (`allocation_count`,`account_id`),
KEY `index_order_lines_on_account_id` (`account_id`),
KEY `index_order_lines_on_allocation_count` (`allocation_count`),
KEY `index_ol_on_acc_id_alloc_count_order_id` (`account_id`,`allocation_count`,`order_id`),
KEY `index_ol_on_order_id_account_id_alloc_count` (`order_id`,`account_id`,`allocation_count`)

order_line_allocations

PRIMARY KEY (`id`),
KEY `index_order_line_allocations_on_account_id` (`account_id`),
KEY `index_order_line_allocations_on_account_type_id` (`account_type_id`),
KEY `index_order_line_allocations_on_order_id` (`order_id`),
KEY `index_order_line_allocations_on_order_line_id` (`order_line_id`)

帐户

PRIMARY KEY (`id`),
KEY `index_accounts_on_account_type_id_and_active` (`account_type_id`,`active`),
KEY `index_accounts_on_segment_1` (`segment_1`),
KEY `index_accounts_on_segment_10` (`segment_10`),
KEY `index_accounts_on_segment_11` (`segment_11`),
KEY `index_accounts_on_segment_12` (`segment_12`),
KEY `index_accounts_on_segment_13` (`segment_13`),
KEY `index_accounts_on_segment_14` (`segment_14`),
KEY `index_accounts_on_segment_15` (`segment_15`),
KEY `index_accounts_on_segment_16` (`segment_16`),
KEY `index_accounts_on_segment_17` (`segment_17`),
KEY `index_accounts_on_segment_18` (`segment_18`),
KEY `index_accounts_on_segment_19` (`segment_19`),
KEY `index_accounts_on_segment_2` (`segment_2`),
KEY `index_accounts_on_segment_20` (`segment_20`),
KEY `index_accounts_on_segment_3` (`segment_3`),
KEY `index_accounts_on_segment_4` (`segment_4`),
KEY `index_accounts_on_segment_5` (`segment_5`),
KEY `index_accounts_on_segment_6` (`segment_6`),
KEY `index_accounts_on_segment_7` (`segment_7`),
KEY `index_accounts_on_segment_8` (`segment_8`),
KEY `index_accounts_on_segment_9` (`segment_9`),
KEY `index_accounts_on_account_type_id` (`account_type_id`)

慢查询的解释计划

*************************** 1. row ***************************
           id: 1
  select_type: SIMPLE
        table: orders
   partitions: NULL
         type: index
possible_keys: index_orders_on_account_type_id,index_orders_on_line_count
          key: PRIMARY
      key_len: 4
          ref: NULL
         rows: 10
     filtered: 100.00
        Extra: NULL
*************************** 2. row ***************************
           id: 1
  select_type: SIMPLE
        table: order_lines
   partitions: NULL
         type: ref
possible_keys: index_order_lines_on_order_id,index_order_lines_on_order_id_and_position,index_order_lines_on_order_id_and_line_num,index_ol_on_header_id_reporting_total_savings_pct_created_at,index_ol_on_order_id_supplier_id_reporting_total,index_ol_on_order_id_commodity_id_reporting_total_ela_id,index_ol_on_order_id_account_id_alloc_count
          key: index_ol_on_order_id_commodity_id_reporting_total_ela_id
      key_len: 4
          ref: perf_amazon_qas1481_utf8mb4.orders.id
         rows: 2
     filtered: 100.00
        Extra: NULL
*************************** 3. row ***************************
           id: 1
  select_type: SIMPLE
        table: order_line_allocations
   partitions: NULL
         type: ref
possible_keys: index_order_line_allocations_on_order_line_id
          key: index_order_line_allocations_on_order_line_id
      key_len: 5
          ref: perf_amazon_qas1481_utf8mb4.order_lines.id
         rows: 3
     filtered: 100.00
        Extra: NULL
*************************** 4. row ***************************
           id: 1
  select_type: SIMPLE
        table: sec_accounts
   partitions: NULL
         type: eq_ref
possible_keys: PRIMARY,index_accounts_on_account_type_id_and_active,index_accounts_on_segment_1,index_accounts_on_account_type_id
          key: PRIMARY
      key_len: 4
          ref: perf_amazon_qas1481_utf8mb4.order_lines.account_id
         rows: 1
     filtered: 100.00
        Extra: Using where
*************************** 5. row ***************************
           id: 1
  select_type: SIMPLE
        table: sec_accounts
   partitions: NULL
         type: eq_ref
possible_keys: PRIMARY,index_accounts_on_account_type_id_and_active,index_accounts_on_segment_1,index_accounts_on_account_type_id
          key: PRIMARY
      key_len: 4
          ref: perf_amazon_qas1481_utf8mb4.order_line_allocations.account_id
         rows: 1
     filtered: 100.00
        Extra: Using where

EXPLAIN PLAN 快速查询：

*************************** 1. row ***************************
           id: 1
  select_type: SIMPLE
        table: orders
   partitions: NULL
         type: index
possible_keys: index_orders_on_account_type_id,index_orders_on_line_count
          key: PRIMARY
      key_len: 4
          ref: NULL
         rows: 10
     filtered: 100.00
        Extra: NULL
*************************** 2. row ***************************
           id: 1
  select_type: SIMPLE
        table: order_lines
   partitions: NULL
         type: ref
possible_keys: index_order_lines_on_order_id,index_order_lines_on_order_id_and_position,index_order_lines_on_order_id_and_line_num,index_ol_on_header_id_reporting_total_savings_pct_created_at,index_ol_on_order_id_supplier_id_reporting_total,index_ol_on_order_id_commodity_id_reporting_total_ela_id,index_ol_on_order_id_account_id_alloc_count
          key: index_ol_on_order_id_commodity_id_reporting_total_ela_id
      key_len: 4
          ref: perf_amazon_qas1481_utf8mb4.orders.id
         rows: 2
     filtered: 100.00
        Extra: NULL
*************************** 3. row ***************************
           id: 1
  select_type: SIMPLE
        table: order_line_allocations
   partitions: NULL
         type: ref
possible_keys: index_order_line_allocations_on_order_line_id
          key: index_order_line_allocations_on_order_line_id
      key_len: 5
          ref: perf_amazon_qas1481_utf8mb4.order_lines.id
         rows: 3
     filtered: 100.00
        Extra: NULL
*************************** 4. row ***************************
           id: 1
  select_type: SIMPLE
        table: sec_accounts
   partitions: NULL
         type: eq_ref
possible_keys: PRIMARY,index_accounts_on_account_type_id_and_active,index_accounts_on_segment_1,index_accounts_on_account_type_id
          key: PRIMARY
      key_len: 4
          ref: perf_amazon_qas1481_utf8mb4.order_lines.account_id
         rows: 1
     filtered: 100.00
        Extra: Using where
*************************** 5. row ***************************
           id: 1
  select_type: SIMPLE
        table: sec_accounts
   partitions: NULL
         type: eq_ref
possible_keys: PRIMARY,index_accounts_on_account_type_id_and_active,index_accounts_on_segment_1,index_accounts_on_account_type_id
          key: PRIMARY
      key_len: 4
          ref: perf_amazon_qas1481_utf8mb4.order_line_allocations.account_id
         rows: 1
     filtered: 100.00
        Extra: Using where

如何加快查询速度？

Answer 1

这些复合索引可能会有所帮助：

order_lines:  INDEX(order_id, account_id, allocation_count)
order_line_allocations:  INDEX(order_line_id,  account_id)
accounts:  INDEX(account_type_id, segment_1)

添加复合索引时，删除具有相同前导列的索引。 也就是说，当您同时拥有INDEX(a)和INDEX(a,b)时，扔掉前者。

可以将两个LEFT JOINs合并到accounts中吗？

如需进一步讨论，请为每个表提供SHOW CREATE TABLE 。 并EXPLAIN 8 和 344。

OR使得使用INDEX进行优化变得不切实际。