MySQL和Oracle中的隱式轉換

lhrbest發表於2017-09-18
今天在處理一個問題的時候,需要根據其他部門提供的sql語句對一個表中的資料進行了篩查。
語句類似下面的形式
> SELECT MAX_LEVEL,LOGOUT_TIME,CURRENT_DATE AS NOWTIME,cn_master FROM t_test_october_back_a WHERE ID in ( 100, 200, 300, 400, 500) ;
+-----------+---------------+------------+-----------+
| MAX_LEVEL | LOGOUT_TIME   | NOWTIME    | ID|
+-----------+---------------+------------+-----------+
|         1 | 1440407918000 | 2015-08-31 | 100|
|        100| 1441009281000 | 2015-08-31 | 200|
|         1 | 1440408002000 | 2015-08-31 | 300|
+-----------+---------------+------------+-----------+
x rows in set, 65535 warnings (10.98 sec)
本來這一個簡單查詢就完成了,也得到了業務部門需要的資料情況,但是檢視最後一行的內容,還是有些蹊蹺。如果觀察仔細,對於這種id的資料查詢,走索引的話,絕對不會再10秒左右,這是第一個奇怪的地方,第二個奇怪的地方就是65535 warnings,一個簡單查詢怎麼會有這麼多的warnings
最開始怕show warnings的時候會一下子顯示出來6萬多行資料,還小小擔心了一下,結果輸出的結果只有64行。
>show warnings;
+---------+------+----------------------------------------------------------------+
| Level   | Code | Message                                                        |
+---------+------+----------------------------------------------------------------+
| Warning | 1292 | Truncated incorrect DOUBLE value: '3506996@abc.com'       |
| Warning | 1292 | Truncated incorrect DOUBLE value: '28366@abc.com'    |
| Warning | 1292 | Truncated incorrect DOUBLE value: '81700700@abc.com'     |
| Warning | 1292 | Truncated incorrect DOUBLE value: '391112900@abc.com'     |
| Warning | 1292 | Truncated incorrect DOUBLE value: '867964771@abc.com'    |
...
64 rows in set (0.00 sec)
檢視建表語句,發現這個id列是varchar型別的。
>show create table  t_test_october_back_a;
| t_tl_october_back_a | CREATE TABLE `t_tl_october_back_a` (
  `id` bigint(20) NOT NULL AUTO_INCREMENT,
  `id` varchar(60) NOT NULL,
  `area_server` varchar(80) NOT NULL,
  `max_level` tinyint(4) NOT NULL,
  `logout_time` bigint(20) NOT NULL,
  PRIMARY KEY (`id`),
  UNIQUE KEY `idx_test_october_back_cn_master` (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=2042252493 DEFAULT CHARSET=gbk |

我們就順著錯誤資訊來看看,把警告中的部分直接作為引數,發現查詢的時間極快。
>select *from t_tl_october_back_a where cn_master ='867964771@abc.com';
+-------+-------------------------+---------------------------+-----------+---------------+
| id    | cn_master               | area_server               | max_level | logout_time   |
+-------+-------------------------+---------------------------+-----------+---------------+
| 18723 | 867964771@abc.com | abcd-abcd            |       104 | 1434446979000 |
+-------+-------------------------+---------------------------+-----------+---------------+
1 row in set (0.00 sec)
對此我們來透過執行計劃來簡單對比一下。
如果使用數字的方式,會走全表掃描
> explain SELECT MAX_LEVEL,LOGOUT_TIME FROM t_tl_october_back_a WHERE CN_MASTER in ( 100, 200);
+----+-------------+---------------------+------+-------------------------------+------+---------+------+----------+-------------+
| id | select_type | table               | type | possible_keys                 | key  | key_len | ref  | rows     | Extra       |
+----+-------------+---------------------+------+-------------------------------+------+---------+------+----------+-------------+
|  1 | SIMPLE      | t_test_october_back_a | ALL  | idx_test_october_back_cn_master | NULL | NULL    | NULL | 28597841 | Using where |
+----+-------------+---------------------+------+-------------------------------+------+---------+------+----------+-------------+
1 row in set (0.00 sec)
而如果使用字元的形式,執行計劃就會走索引,執行效果就是預期的樣子。
>explain SELECT MAX_LEVEL,LOGOUT_TIME FROM t_tl_october_back_a WHERE CN_MASTER = '100' ;
+----+-------------+---------------------+-------+-------------------------------+-------------------------------+---------+-------+------+-------+
| id | select_type | table               | type  | possible_keys                 | key                           | key_len | ref   | rows | Extra |
+----+-------------+---------------------+-------+-------------------------------+-------------------------------+---------+-------+------+-------+
|  1 | SIMPLE      | t_test_october_back_a | const | idx_test_october_back_cn_master | idx_test_october_back_cn_master | 122     | const |    1 |       |
+----+-------------+---------------------+-------+-------------------------------+-------------------------------+---------+-------+------+-------+
1 row in set (0.00 sec)
透過上面的例子可以看到,在查詢的時候丟擲的警告,其實就是在做型別轉換的時候本來輸出的是數字型別,就會嘗試做隱式轉換,而那個65535只是一個最大限制而已,表中的資料其實已經遠遠超過千萬。
這個時候我們大體感受到了隱式轉換在MySQL中的一些影響,我們來看看在MySQL 5.6和Oracle中的表現如何。
###############
MYSQL 5.6
mysql> select version();
+-------------------------------------------+
| version()                                 |
+-------------------------------------------+
| 5.6.23-enterprise-commercial-advanced-log |
+-------------------------------------------+
mysql> create table test (id1 int,id2 varchar(10));
mysql> insert into test values(1,'1');
mysql> insert into test values(2,'2');
mysql> insert into test values(3,'3');
mysql> commit;
mysql> create index idx_id1 on test(id1);
mysql> create index idx_id2 on test(id2);
隱式轉換,有數字轉換為字元的時候,直接走了索引掃描
mysql> explain select * from test where id1='1';
+----+-------------+-------+------+---------------+---------+---------+-------+------+-------+
| id | select_type | table | type | possible_keys | key     | key_len | ref   | rows | Extra |
+----+-------------+-------+------+---------------+---------+---------+-------+------+-------+
|  1 | SIMPLE      | test  | ref  | idx_id1       | idx_id1 | 5       | const |    1 | NULL  |
+----+-------------+-------+------+---------------+---------+---------+-------+------+-------+
1 row in set (0.01 sec)

隱式轉換,由字元轉換為數字的時候,直接走了全表掃描
mysql> explain select *from test where id2=2;
+----+-------------+-------+------+---------------+------+---------+------+------+-------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows | Extra       |
+----+-------------+-------+------+---------------+------+---------+------+------+-------------+
|  1 | SIMPLE      | test  | ALL  | idx_id2       | NULL | NULL    | NULL |    3 | Using where |
+----+-------------+-------+------+---------------+------+---------+------+------+-------------+
1 row in set (0.00 sec)


###############
Oracle 11gR2

sqlplus -v
SQL*Plus: Release 10.2.0.3.0 - Production
SQL> create table test (id1 number,id2 varchar2(10));
SQL> insert into test values(1,'1');
SQL> begin                   
  2  for i in 1..100 loop
  3  insert into test values(i,chr(39)||i||chr(39));
  4  end loop;
  5  commit;
  6  end;
  7  /
PL/SQL procedure successfully completed.
SQL> exec dbms_stats.gather_table_stats('TEST','TEST',CASCADE=>TRUE);
隱式轉換,由數字轉換為字元的時候,直接走了索引掃描
SQL> explain plan for select *from test where id1='2';
SQL> select *from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Plan hash value: 3847161316
---------------------------------------------------------------------------------------
| Id  | Operation                   | Name    | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |         |     1 |     7 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| TEST    |     1 |     7 |     2   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | IDX_ID1 |     1 |       |     1   (0)| 00:00:01 |
---------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
PLAN_TABLE_OUTPUT
   2 - access("ID1"=2)

隱式轉換,由字元轉換為數字的時候,直接走了全表掃描
SQL> explain plan for select *from test where id2=3;
SQL> select *from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Plan hash value: 1357081020
--------------------------------------------------------------------------
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |      |     1 |     7 |     3   (0)| 00:00:01 |
|*  1 |  TABLE ACCESS FULL| TEST |     1 |     7 |     3   (0)| 00:00:01 |
--------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
   1 - filter(TO_NUMBER("ID2")=3)

可見在這個方面MySQL和Oracle中的表現是一致的,對於這種隱式轉換還是要多加註意。


來自 “ ITPUB部落格 ” ,連結:http://blog.itpub.net/26736162/viewspace-2145096/,如需轉載,請註明出處,否則將追究法律責任。

相關文章