Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Should I use prepared statements for MySQL in PHP PERFORMANCE-WISE?

Tags:

I understand the security benefits of prepared statements in MySQL. No need to cover that topic here. I'm wondering about the performance aspect of them.

Now, I know when a query using a prepared statement is executed twice in a single PHP script, it's faster because the query is only parsed once, once for each query. The client makes one trip to prepare, then sends data twice using the binary protocol. The binary protocol is faster, and you're not taking the hit of having to parse a second time.

However, what about the case where I only want to perform a query once in a single PHP script? It would seem using a prepared statement is worse, because you're making two trips to the server, once to prepare, and once to send the data. The benefit of only having to parse once is lost, and you're penalized for that second trip. If the data isn't sufficiently smaller in binary format, you lose by using a prepared statement, right?

However, I've read some conflicting reports about what PHP's mysqli or PDO libraries do? Do either of them cache the prepared statement across script execution? Is the server going to have to parse the prepared statement again on a subsequent pageload or not? If the answer is no, that the statement doesn't have to be parsed on the second pageload, then it would seem that prepared statements ARE better, even if you're only executing the query once per pageload.

Please take into consideration if anything has changed between versions of MySQL regarding this. You can safely assume I'm using PHP 5.2

EDIT: Just to make it clear, I want an answer for MySQL and PHP specifically, specifying the MySQL version and if this was ever different, and to ONLY consider performance, not ease of use or security.

UPDATE: I accepted the answer I did because of the follow up comment had a few good ideas. I'm still a bit disappointed that no one seems to be able to answer the crux of the actual question I asked with any certainty. I guess sometimes the answer really is "it depends."

like image 428
Mike Sherov Avatar asked Feb 06 '10 19:02

Mike Sherov


People also ask

Should I use PHP prepared statements?

You must always use prepared statements for any SQL query that would contain a PHP variable.

Is PreparedStatement faster?

Prepared statements are much faster when you have to run the same statement multiple times, with different data. Thats because SQL will validate the query only once, whereas if you just use a statement it will validate the query each time.

Should I always use prepared statements?

You should always prefer working with prepared statements for the security benefits. They all but eliminate vulnerability to SQL injection, without you having to worry about SQL-escaping values. If you have a query that doesn't run often, though (less than once per request), a prepared statement can take longer to run.

Are prepared statements slower?

Prepared statements are generally faster than regular queries if you're repeatedly running the same query. Performance is nice but the real win with PreparedStatements is the parameter binding can do done via the API rather than string concatenation.


2 Answers

The History

This was my first Stackoverflow answer. A lot has changed since, specially the deprecation and removal of the mysql API. Even if you are still on php 5.6, the mysql_* api should not be used. Now PDO or mysqli are the only options to choose. PDO is better to lots of reasons.

Are prepared statements cached across page loads?

I've read some conflicting reports about what PHP's mysqli or PDO libraries do? Do either of them cache the prepared statement across script execution?

The same prepared statement will not be used in between page loads. It has to be prepared every time. If squeezing every large millisecond matters, a stored procedure might be a good idea (assuming you have a complicated query).

For large inserts (thousands of rows) A bigger boost can probably be gained by dumping your data into a text file and loading it with LOAD DATA IN FILE . It's a lot faster than a series of inserts.

The original answer

The truth of the matter is that sometimes mysqli is faster and at other times mysql api is faster. But the difference is really really small. If you look at any of the performance tests on the web the difference is really just 10 - 20 milliseconds. The best way to boost performance is to optimize table design.

Many of the tests that 'prove' the older api to be faster conveniently forget that for maximum security mysql_real_escape_string() should be called for each variable used in the query.

Queries are cached by the server, if and only if the data on all the tables that are used in the query have remained unchanged.

Await another update with actual numbers

like image 140
e4c5 Avatar answered Oct 09 '22 11:10

e4c5


"Using prepared statements is never really a bad idea"

Not true. You can easily fill up the prepared statement cache and exceed max_prepared_stmt_count if you don't know what you're doing, rendering your app useless until connections consuming the prepared statements are closed.

Prepared statements are specifically designed for tight inner loops in your business logic where you're going to be calling the same basic query over and over again, a good example being a parameterized query such as:

SELECT name, address, phone from tbl WHERE id = ? 

and you have a different id on each call. That way, it's worth the extra round trip to the db for the prepare because you're probably going to call it hundreds or thousands of times and just change the parameter. But you're expected to remove the prepared statement from the cache or close the connection at the end of, say, your stateless script (php, perl, jsp, ruby, etc).
If you don't remove the prepared statement and you are using a connection pool, you are bound to fill up the cache over time and get a nasty error "Can't create more than max_prepared_stmt_count statements". I'm speaking from experience so think about whether you really need prepared statements because you know definitively that you're going to be repeating the same parameterized query over and over in a tight loop.
If not, what you're probably looking for is letting mysql use its basic query cache which is a different mechanism than the prepared statement list and which behaves like a true LRU cache from my understanding.

like image 24
NormK Avatar answered Oct 09 '22 10:10

NormK