Question

For example if I have a table User, I want to store gender or sex, I’ll add a column like sex.

Is it really worth to use an integer and then map it in my favorite programming language?

Like 1 => 'Male' and 2 => 'Female'

Is there any performance reason to do that?

Or could I safely use a varchar which more meaning with ‘female’ or ‘male’ almost like I was using mysql ENUM ?

Edit: I here and there that it is sometimes better, sometimes it doesn’t matter, so I more looking for benchmark or something over a “it is better” answer.

I mean I think using varchar is actually more meaningfull than an integer, and I would use an integer only if performance are more than 0.3% or something.

Answer 1

Ortiginal Answer:
I would suggest storing it in a CHAR(1) column as M or F
It is expressive enough for the specific purpose AND has the speed benefit of being a single character comparison

Update 4 (fixed benchmark):
All previous benchmarks had a fatal flaw that one (the CHAR(1)) table was MyISAM and all other were InnoDB. So I recreated the database with all tables using the MyISAM and the results make much more sense now.

The error creeped in as I used the MySQLWorkbench’s wizard to create the tables and forgot to change the database engine in the other tables and it defaulted to InnoDB (I have MySQL 5.5)

So the corrected results are as follows, (I have removed all my previous benchmarks as they were invalid) :

// select queries
$query['char'] = "select count(*) from test_table where gender = 'M'";
$query['char_lower'] = "select count(*) from test_table where LOWER(gender) = 'm'";
$query['varchar'] = "select count(*) from test_table_2 where gender = 'Male'";
$query['varchar_lower'] = "select count(*) from test_table_2 where LOWER(gender) = 'male'";
$query['tinyint'] = "select count(*) from test_table_3 where gender = 1";

// benchmark result
array
  'char' => float 0.35457420349121
  'char_lower' => float 0.44702696800232
  'varchar' => float 0.50844311714172
  'varchar_lower' => float 0.64412498474121
  'tinyint' => float 0.26296806335449

New Conclusion : TINYINT Is fastest. But my recommendation would be still yo use CHAR(1) as it would be easier for future developers to understand the database.

If you do use TINYINT, my recommendation would be name the column ismale instead of sex and store 0 => Female and 1 => male thus making it a little more easy to understand in raw database.

The table structure for benchmark is this:

CREATE TABLE `test_table` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `gender` char(1) DEFAULT NULL,
  PRIMARY KEY (`id`)
) ENGINE=MyISAM

Only the type of the gender column is different in the 3 tables, the types are:

CHAR(1), VARCHAR(6), TINYINT

All 3 tables have 10000 entries.