[MySQL優化案例]系列 -- 用TIMESTAMP類型取代INT和DATETIME

時間 2021-08-13

標籤 mysql sql ide 測試 this spa unix 索引 get 博客欄目 MySQL 简体版

原文原文鏈接

做/譯者：葉金榮（Email:

），來源：http://imysql.cn，轉載請註明做/譯者和出處，而且不能用於商業用途，違者必究。

引言：在之前，我老是習慣用 INT UNSIGNED 來存儲一個轉換成Unix時間戳的時間值，認爲這樣作從索引，比較等角度來說，都會比較高效。如今咱們來對比下 TIMESTAMP 和 INT UNSIGNED 以及 DATETIME 這3種類型到底誰更好。

1. 準備

建立一個測試表：

mysql> CREATE TABLE `t` (
`d1` int(10) unsigned NOT NULL default '0',
`d2` timestamp NOT NULL default CURRENT_TIMESTAMP,
`d3` datetime NOT NULL,
KEY `d2` (`d2`),
KEY `d1` (`d1`),
KEY `d3` (`d3`)
);

而後建立一個存儲過程填充數據：

mysql> DELIMITER //
CREATE PROCEDURE INS_T()
BEGIN
SET @i=1;
WHILE 0<1
DO
SET @i=@i+1;
INSERT INTO i VALUES (1199116800+@i, FROM_UNIXTIME(1199116800+@i), FROM_UNIXTIME(1199116800+@i));
END WHILE;
END;//
DELIMITER ;

時間戳 1199116800 表示 2008-01-01 這個時間點。而後運行存儲過程，大概填充幾十萬條記錄後，停止執行，由於上面的存儲過程是個死循環，因此須要人工停止。
來看看到底有多少條記錄了，以及索引狀況：

mysql> select count(*) from t;
+----------+
| count(*) |
+----------+
|   924707 |
+----------+
mysql> analyze table t;
+--------+---------+----------+-----------------------------+
| Table  | Op      | Msg_type | Msg_text                    |
+--------+---------+----------+-----------------------------+
| test.t | analyze | status   | Table is already up to date |
+--------+---------+----------+-----------------------------+
mysql> show index from t;
+-------+------------+----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment |
+-------+------------+----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+
| t     |          1 | d2       |            1 | d2          | A         |      924707 |     NULL | NULL   |      | BTREE      |         |
| t     |          1 | d1       |            1 | d1          | A         |      924707 |     NULL | NULL   |      | BTREE      |         |
| t     |          1 | d3       |            1 | d3          | A         |      924707 |     NULL | NULL   |      | BTREE      |         |
+-------+------------+----------+--------------+-------------+-----------+-------------+----------+--------+------+------------+---------+

2. 對比

2.1 只檢索一條記錄

mysql> explain select * from t where d1 = 1199579155;
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
| id | select_type | table | type | possible_keys | key  | key_len | ref   | rows | Extra |
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
|  1 | SIMPLE      | t     | ref  | d1            | d1   | 4       | const |    1 |       |
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
mysql> explain select * from t where d2 = '2008-01-06 08:25:55';
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
| id | select_type | table | type | possible_keys | key  | key_len | ref   | rows | Extra |
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
|  1 | SIMPLE      | t     | ref  | d2            | d2   | 4       | const |    1 |       |
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
mysql> explain select * from t where d3 = '2008-01-06 08:25:55';
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
| id | select_type | table | type | possible_keys | key  | key_len | ref   | rows | Extra |
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+
|  1 | SIMPLE      | t     | ref  | d3            | d3   | 8       | const |    1 |       |
+----+-------------+-------+------+---------------+------+---------+-------+------+-------+

2.2 範圍檢索

mysql> explain select * from t where d1 >= 1199894400;
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
| id | select_type | table | type  | possible_keys | key  | key_len | ref  | rows   | Extra       |
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
|  1 | SIMPLE      | t     | range | d1            | d1   | 4       | NULL | 121961 | Using where |
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
mysql> explain select * from t where d2 >= from_unixtime(1199894400);
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
| id | select_type | table | type  | possible_keys | key  | key_len | ref  | rows   | Extra       |
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
|  1 | SIMPLE      | t     | range | d2            | d2   | 4       | NULL | 121961 | Using where |
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
mysql> explain select * from t where d3 >= from_unixtime(1199894400);
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
| id | select_type | table | type  | possible_keys | key  | key_len | ref  | rows   | Extra       |
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+
|  1 | SIMPLE      | t     | range | d3            | d3   | 8       | NULL | 120625 | Using where |
+----+-------------+-------+-------+---------------+------+---------+------+--------+-------------+

小結：從上面的2次對比中能夠看到，對 d1 或 d2 字段檢索時的索引長度都是 4，由於 TIMESTAMP 其實是 4字節 的 INT 值。所以，實際應用中，基本上徹底能夠採用 TIMESTAMP 來代替另外2種類型了，而且 TIMESTAMP 還能支持自動更新成當前最新時間，何樂而不爲呢？