I’m in a dilemma about saving date and time values in MySQL’s TIMESTAMP format vs in a custom UNSIGNED INT format. The main considerations here are speed of retrieval, appropriate range calculations in PHP and occasional formatting into human readable values.
The storage space required for each type and their ranges:
DATETIME 8 bytes '1000-01-01 00:00:00' to '9999-12-31 23:59:59' TIMESTAMP 4 bytes '1970-01-01 00:00:01' UTC to '2038-01-19 03:14:07' UTC UNSIGNED INT 4 bytes (Maximum Value 4294967295)
I dont need the range of DATETIME at all. I’m torn between TIMESTAMP and UNSIGNED INT.
Arguments in favor of UNSIGNED INT:
- A UNIX timestamp of 4294967295 converts to Sun, 07 Feb 2106 06:28:15 GMT which is more than TIMESTAMP and good enough for me
- Comparing these timestamps directly in PHP would be faster rather than converting TIMESTAMPs via strtotime() and then comparing them
The only advantage TIMESTAMP would give me is when I’m reading in the values from the mysql table manually and need to ‘see’ them.
Is there any compelling reason to use TIMESTAMP and not an UNSIGNED INT?
Arguments for TIMESTAMP
- It implicitly stores data in GMT time zone. No matter what your session time-zone is. Useful if you need to use different timezones.
- You can have automated timestamping columns using
ON UPDATE CURRENT_TIMESTAMP(one column per table only until MySQL 5.6.5)
- You can use datetime function for date comparision, addition, substraction, range lookup etc, without the need to use
FROM_UNIXTIME()function – it will make it easier to write queries that can use indexes
>> date('Y-m-d h:i:s',4294967295); '1969-12-31 11:59:59'
so the range is in fact the same
- You can still retrieve integer unix timestamp with no additional overhead using UNIX_TIMESTAMP() function: http://dev.mysql.com/doc/refman/5.5/en/date-and-time-functions.html#function_unix-timestamp
When UNIX_TIMESTAMP() is used on a TIMESTAMP column, the function
returns the internal timestamp value directly, with no implicit
The only real use for TIMESTAMP is when you want that field to be updated automatically when the row is updated (which is the default behaviour for that field), or when data storage requirements are so strict that 4 bytes per row really makes a difference to you.
Really the comparison should be between DATETIME and UNSIGNED INT, and I’d recommend DATETIME because:
- You can use MySQL’s native date/time functions for selecting by date ranges etc.
- It is trivially easy to select these dates out as UNIX timestamps for easy formatting in PHP:
SELECT UNIX_TIMESTAMP(field) FROM table, no need to select out the raw value and use strtotime
- Easier to read and edit the fields in your database directly if you need to (as you pointed out).
- No limitations on date range
Point two alone really removes any reason to store in integers, in my opinion.
This might not be a “scientific” answer but I always find the way MySql handles conversion, arithmetics, comparsion, etc… on TIMESTAMP columns confusing. An UNSIGNED INT column is much more straight forward and I always know what to expect.
P.S. Perhaps one other thing in favor of TIMESTAMP column is its ability to be automatically set to current time after each update or insert but that is not something you can’t live without.