如何使用索引表MariaDB改善慢速查詢？

Question

該查詢平均需要2/3秒。 聯接中的每個字段以及索引在哪里。

我如何改善此查詢？

SELECT DISTINCT pro_id, pro_url, pro_data, pro_capa, pro_destaque, pro_destaque_data, pro_nome, pro_likes, pro_views, pro_comentarios_total, pro_autor FROM projeto JOIN utilizador_projeto ON pro_id=utp_proid JOIN utilizador ON utp_utiid=uti_id WHERE pro_activo=1 AND pro_privacidade=1 ORDER BY pro_destaque_data DESC LIMIT 24;

slow_query_log：

# Time: 190923  1:10:58
# User@Host: root[root] @  [10.133.247.241]
# Thread_id: 36  Schema: db  QC_hit: No
# Query_time: 3.575462  Lock_time: 0.000114  Rows_sent: 24  Rows_examined: 104820
# Rows_affected: 0  Bytes_sent: 3974

MariaDB [db]> EXPLAIN SELECT DISTINCT pro_id, pro_url, pro_data, pro_capa, pro_destaque, pro_destaque_data, pro_nome, pro_likes, pro_views, pro_comentarios_total, pro_autor FROM projeto JOIN utilizador_projeto ON pro_id=utp_proid JOIN utilizador ON utp_utiid=uti_id WHERE pro_activo=1 AND pro_privacidade=1 ORDER BY pro_destaque_data DESC LIMIT 24;
+------+-------------+--------------------+--------+-----------------------------------------------------------------------------------------------+---------------------+---------+---------------------------------+-------+------------------------------+
| id   | select_type | table              | type   | possible_keys                                                                                 | key                 | key_len | ref                             | rows  | Extra                        |
+------+-------------+--------------------+--------+-----------------------------------------------------------------------------------------------+---------------------+---------+---------------------------------+-------+------------------------------+
|    1 | SIMPLE      | projeto            | ref    | PRIMARY,pro_destaques_index,pro_likes_index,pro_comments_index,pro_views_index,pro_date_index | pro_destaques_index | 2       | const,const                     | 17865 | Using where; Using temporary |
|    1 | SIMPLE      | utilizador_projeto | ref    | utp_utiid,utp_proid                                                                           | utp_proid           | 4       | db.projeto.pro_id               |     1 | Distinct                     |
|    1 | SIMPLE      | utilizador         | eq_ref | PRIMARY                                                                                       | PRIMARY             | 4       | db.utilizador_projeto.utp_utiid |     1 | Using index; Distinct        |
+------+-------------+--------------------+--------+-----------------------------------------------------------------------------------------------+---------------------+---------+---------------------------------+-------+------------------------------+

Answer 1

    WHERE  pro_active=1
      AND  pro_privacy=1
    ORDER BY  pro_highlight_date DESC

乞求這個“復合”索引：

INDEX(pro_active, pro_privacy,   -- in either order
      pro_highlight_date)

Rows_examined: 139482可能會降為Rows_examined: 24 。

假定這3列在同一表中。 請執行以下操作，而不用表名作為列名的前綴：

...
FROM project AS pro
...
WHERE pro.active = 1 ...

這樣就消除了任何歧義（並且擊鍵次數大致相同）。

聯接中的每個字段以及索引在哪里

過度殺傷力。 而且，單列索引通常不如多列（復合）索引有用。 參見： http : //mysql.rjweb.org/doc.php/index_cookbook_mysql