在MySQL和php中使用文本与整数和应用内常量“类型”列之间的性能差异在哪里？

Question

In my case, I have a table which stores a collection of records with similar information but each with unique type column, used in various parts of my application. I know, I know this is "micro-optimisation" but it it an integral part of my application (it will store many records), and I would like it to be optimised, and I am also simply curious, is is faster to use text type and select it like
SELECT ... WHERE type = 'some_type' or use a PHP defined constant like const SOMETYPE = 1; run_query('SELECT ... WHERE type = '.SOMETYPE); ?

Answer 1

String comparison will always be slower than integer comparison. Typically, what is done is the string are stored in a separate table, perhaps called standard_types or whatever makes sense for the "constants" being stored. That table then has a unique id field that can be referenced by other tables.

This way if you need the strings for reporting, the reporting queries can join to the "types" table for the display strings. Ideally, in my opinion, the id values should reflect a standard numbers that can be expressed as enum values or constants in client code; this minimizes the dependence on the "types" table for non-reporting queries.

Some might argue against keeping the list of standard id values and their meanings coordinated across the database and one or more application codebases; but the alternative is coordinating standard strings across all that (domains that can handle those string values quite differently).

Answer 2

The dominant time spent in any query is in fetching rows to work with. Functions, string vs int, etc, make only minor differences in performance. Focus on what is clean for your code, and what minimizes the number of rows touched.

Once you have done that, minimize the number of round trips to the server. Even so, I have created many web pages that do 20-50 queries (each well optimized); the page performance is adequate.

You could also consider the ENUM data type.

sex ENUM('unk', 'male', 'female') NOT NULL

gives you WHERE sex = 'male' , implemented as a 1-byte integer under the covers.