the rule of thumb is…

1. use count(primary_key)
2. use count(indexed_field_and_not_null)
3. use count(indexed_field)
4. use count(*)

how abt the length of the index, does it matter?

thanks.

For primary key it should be the same as primary key can’t have null values.
It is same as count(val2) in my example.

Using innodb + where and myisam + where;

Good point. It looks like a bug to me as Innodb automatically picks up “shorter” indexes while in practice even first index you tried is shorter.

The other interesting point I should note - smaller index may not be faster for huge tables, because if you insert data in order primary key tree is often sequential while secondary indexes may be scattered having few sequential pages.

Consider this example (restarting mysql each time to flush the cache, and using O_DIRECT to bypass OS caching)

mysql> explain select count(*) from wp_comments;
+—-+————-+————-+——-+—————+———+———+——+——–+————-+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+—-+————-+————-+——-+—————+———+———+——+——–+————-+
| 1 | SIMPLE | wp_comments | index | NULL | PRIMARY | 8 | NULL | 424672 | Using index |
+—-+————-+————-+——-+—————+———+———+——+——–+————-+
1 row in set (0.11 sec)

mysql> explain select count(rajid) from wp_comments;
+—-+————-+————-+——-+—————+——–+———+——+——–+————-+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+—-+————-+————-+——-+—————+——–+———+——+——–+————-+
| 1 | SIMPLE | wp_comments | index | NULL | rajidx | 9 | NULL | 424672 | Using index |
+—-+————-+————-+——-+—————+——–+———+——+——–+————-+
1 row in set (0.00 sec)

mysql> select count(*) from wp_comments;
+———-+
| count(*) |
+———-+
| 230120 |
+———-+
1 row in set (32.57 sec)

and if you look at the buffer pool used for the count(*):

Total memory allocated 589194920; in additional pool allocated 714496
Buffer pool size 32768
Free buffers 14104
Database pages 18664
Modified db pages 0

restart mysql and check the count of (rajid):

mysql> select count(rajid) from wp_comments;
+————–+
| count(rajid) |
+————–+
| 230120 |
+————–+
1 row in set (0.28 sec)

Much faster. Why? Because it only reads the index data, not the row data for the table when you use a secondary index. Check the buffer pool usage:

Total memory allocated 589194920; in additional pool allocated 714496
Buffer pool size 32768
Free buffers 32332
Database pages 436
Modified db pages 0

The column rajid btw, is just a copy of the PK column comment_ID with a unique index:

mysql> show create table wp_comments\G
*************************** 1. row ***************************
Table: wp_comments
Create Table: CREATE TABLE `wp_comments` (
`comment_ID` bigint(20) unsigned NOT NULL auto_increment,
`comment_post_ID` int(11) NOT NULL default ‘0′,
`comment_author` tinytext NOT NULL,
`comment_author_email` varchar(100) NOT NULL default ”,
`comment_author_url` varchar(200) NOT NULL default ”,
`comment_author_IP` varchar(100) NOT NULL default ”,
`comment_date` datetime NOT NULL default ‘0000-00-00 00:00:00′,
`comment_date_gmt` datetime NOT NULL default ‘0000-00-00 00:00:00′,
`comment_content` text NOT NULL,
`comment_karma` int(11) NOT NULL default ‘0′,
`comment_approved` enum(’0′,’1′,’spam’) NOT NULL default ‘1′,
`comment_agent` varchar(255) NOT NULL default ”,
`comment_type` varchar(20) NOT NULL default ”,
`comment_parent` bigint(20) NOT NULL default ‘0′,
`user_id` bigint(20) NOT NULL default ‘0′,
`rajid` bigint(20) unsigned default NULL,
PRIMARY KEY (`comment_ID`),
UNIQUE KEY `rajidx` (`rajid`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1
1 row in set (0.00 sec)

mysql>

btw, if there is a smaller secondary index on the table, innodb is smart enough to use that for a count, even if it is not a unique index. For instance:
mysql> explain select count(*) from wp_comments;
+—-+————-+————-+——-+—————+———————-+———+——+——–+————-+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+—-+————-+————-+——-+—————+———————-+———+——+——–+————-+
| 1 | SIMPLE | wp_comments | index | NULL | comment_approved_idx | 1 | NULL | 228484 | Using index |
+—-+————-+————-+——-+—————+———————-+———+——+——–+————-+
1 row in set (0.23 sec)

mysql> select count(*) from wp_comments;
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect…
Connection id: 2
Current database: test

+———-+
| count(*) |
+———-+
| 230120 |
+———-+
1 row in set (1.19 sec)

mysql>
and the buffer pool usage:

Total memory allocated 589194920; in additional pool allocated 721920
Buffer pool size 32768
Free buffers 32449
Database pages 319
Modified db pages 0

there’s only 3 distinct values in the table for this column, so its not actually counting the values but the pointers from this index:
mysql> select count( distinct (comment_approved) ) from wp_comments;
+————————————–+
| count( distinct (comment_approved) ) |
+————————————–+
| 3 |
+————————————–+
1 row in set (0.32 sec)

mysql>

cheers!
–Raj.

mysql> explain extended select count(*) from mytable;
+—-+————-+——-+——+—————+——+———+——+——+——————————+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+—-+————-+——-+——+—————+——+———+——+——+——————————+
| 1 | SIMPLE | NULL | NULL | NULL | NULL | NULL | NULL | NULL | Select tables optimized away |
+—-+————-+——-+——+—————+——+———+——+——+——————————+
1 row in set, 1 warning (0.00 sec)

mysql> show warnings;
+——-+——+———————————————————+
| Level | Code | Message |
+——-+——+———————————————————+
| Note | 1003 | select count(0) AS `count(*)` from `database`.`mytable` |
+——-+——+———————————————————+
1 row in set (0.00 sec)

count(1) is similar to count(1) as you can se here:

mysql> explain extended select count(1) from mytable;
+—-+————-+——-+——+—————+——+———+——+——+——————————+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+—-+————-+——-+——+—————+——+———+——+——+——————————+
| 1 | SIMPLE | NULL | NULL | NULL | NULL | NULL | NULL | NULL | Select tables optimized away |
+—-+————-+——-+——+—————+——+———+——+——+——————————+
1 row in set, 1 warning (0.00 sec)

mysql> show warnings;
+——-+——+———————————————————+
| Level | Code | Message |
+——-+——+———————————————————+
| Note | 1003 | select count(1) AS `count(1)` from `database`.`mytable` |
+——-+——+———————————————————+
1 row in set (0.00 sec)

So in conclusion count(*) == count(0) and count(*) != count(1)

Hope it helps.