mysql - 如何优化查询以删除重复的 mySQL？

Question

我在 Google Cloud 上的 mySQL 数据库中有一个包含 433,333 条记录的表，如下所示:

Album_ID | Track_Len | Track_Name                            | Ft_LName1 | Ft_FName1 | Ft_LName2 | Ft_FName2 | Ft_LName3 | Ft_FName3 | Row_Num |
+---------+-----------+---------------------------------------+-----------+-----------+-----------+-----------+-----------+-----------+---------+
| N40781  |      5.19 | Tumbala (Da Lata Remix)               | NULL      | Novalima  | NULL      | NULL      | NULL      | NULL      |       1 |
| N40781  |      5.01 | Ruperta (Zeb Remix)                   | NULL      | Novalima  | NULL      | NULL      | NULL      | NULL      |       2 |
| N40781  |      6.35 | Coba Guarango (Toni Economides Remix) | NULL      | Novalima  | NULL      | NULL      | NULL      | NULL      |       3 |
| B15033  |      6.02 | II-V-P                                | Quartet   | ARC       | NULL      | NULL      | NULL      | NULL      |       4 |
| N32395  |      4.47 | My Babe                               | Stigers   | Curtis    | NULL      | NULL      | NULL      | NULL      |       5 |
| N32395  |      5.13 | Thats All Right                       | Stigers   | Curtis    | NULL      | NULL      | NULL      | NULL      |       6 |

注意主键应该是 (Album_ID,Track_Name)。 有很多重复项，所以我正在运行以下命令来尝试消除它们:

delete from Track where (Album_ID, Track_Name, Row_Num) IN(
  select Album_ID, Track_Name, MAX(Row_Num)
  from (select Album_ID,Track_Name,Row_Num from Track) as x
  where (Album_ID, Track_Name) IN(
                  select Album_ID,Track_Name
                  from (select Album_ID,Track_Name from Track) as y
                  group by Album_ID, Track_Name
                  having count(*) > 1
                )
  group by Album_ID,Track_Name);

但这花费的时间太长，并且不会立即删除所有重复项。 有什么优化此查询的建议吗？？

+------------+---------------------+------+-----+---------+----------------+
| Field      | Type                | Null | Key | Default | Extra          |
+------------+---------------------+------+-----+---------+----------------+
| Album_ID   | varchar(6)          | YES  |     | NULL    |                |
| Track_Len  | decimal(4,2)        | YES  |     | NULL    |                |
| Track_Name | varchar(100)        | YES  |     | NULL    |                |
| Ft_LName1  | varchar(40)         | YES  |     | NULL    |                |
| Ft_FName1  | varchar(40)         | YES  |     | NULL    |                |
| Ft_LName2  | varchar(40)         | YES  |     | NULL    |                |
| Ft_FName2  | varchar(40)         | YES  |     | NULL    |                |
| Ft_LName3  | varchar(40)         | YES  |     | NULL    |                |
| Ft_FName3  | varchar(40)         | YES  |     | NULL    |                |
| Row_Num    | bigint(20) unsigned | NO   | PRI | NULL    | auto_increment |
+------------+---------------------+------+-----+---------+----------------+

Answer 1

在 MySQL 中执行此操作的传统方法是使用 JOIN 和 GROUP BY:

delete t
    from Track t left join
         (select tt.Album_ID, tt.Track_Name, min(tt.row_num) as min_row_num
          from Track tt
          group by tt.Album_ID, tt.Track_Name
         ) tt
         on t.row_number = tt.min_row_num
    where tt.min_row_num is null;

这利用了 id 整体上唯一且可能是主键的事实。您还可以将其指定为:

delete t
    from Track t join
         (select tt.Album_ID, tt.Track_Name, min(tt.row_num) as min_row_num
          from Track tt
          group by tt.Album_ID, tt.Track_Name
         ) tt
         on tt.Album_ID = t.Album_ID and
            tt.Track_Name = t.Track_Name and
            t.row_number > tt.min_row_num;