5 Secrets to Better PHP Performance

December 02 2014
 

Here are 5 quick tricks to get better PHP application performance. Read More


Wait! Do you really need to profile that PHP code? Are you sure you want to start down that time-consuming, tedious path? If you’re looking to squeeze some more performance out of your PHP web application, there are a few relatively quick and easy checks to perform that can give your performance a boost before you dive into refactoring the code. And even if you’re intent on profiling your PHP code, you should still look at these areas to make sure you’re getting maximum performance.

Cache In On OPCache

One of PHP’s strengths is that it’s interpreted on the fly into executable code called opcodes, so you can develop rapidly and test without pausing to compile your code with every change you make. However, it’s inefficient and slow to recompile the identical code each time that code runs on your website.

For many years, opcode cache has been a go-to solution for this particular slowdown. These caches are PHP extensions that create hooks into the compilation system and save the output of compiled code into memory. Then in future runs, PHP checks to make sure that the source file has not changed — via timestamps and file size checking — and if it hasn’t, it runs the cached copy of the code.

The most famous of these caches was APC or Alternative PHP Cache. APC not only provided an opcode cache, but also permitted making user data persistent in shared memory as well.

Given the importance of having an opcode cache configured to get optimal performance out of a PHP application, the PHP core team decided to include one by default with all versions of PHP since version 5.5. They chose OPCache, formerly Zend Optimizer+. Part of the commercial Zend Server offering, Zend Optimizer has now been open-sourced back to the community.

(To learn more about the importance of OpCache to PHP application performance, see this excellent article by my colleague Rob Bolton.)

Screen Shot 2014-12-01 at 9.07.55 AM

Look Outside Your Application

Maybe it’s not your PHP at all that’s slowing your application down, especially if you’ve implemented opcode cache. It’s more than likely that at least several of your application bottlenecks happen when accessing external resources. Let’s look at a couple suspects.

Database Delays

It’s not untypical for the database layer to account for 90 percent of measured execution time for a PHP application. So it makes sense to spend the necessary time to review your codebase for all instances of database access.

First and most obvious, turn on the slow SQL logs and find and fix slow SQL queries. Then proceed to query the queries themselves. Are they efficient? Do you make the same queries multiple times in one execution of your code? (Even with a query cache, that’s still inefficient.) Are you making too many queries? Do you have queries hitting a table without an appropriate index?
Investing a little time to fix your queries can noticeably reduce your database access time and noticeably increase your application performance.

Filesystem Snafus

I/O, I/O, who knows where the time goes? Some of it is with all the in-and-out of your file system. So study your filesystem for the same kinds of inefficiencies you looked for in your database queries. Some likely time-consuming culprits: reading in local files, processing XML, image processing, or using the filesystem for session storage.

Specifically, look for code that would cause a file stat to happen — reading of a file’s statistics, such as the date it was last modified. Functions such as file_exists(), filesize(), or filetime() cause file stats to happen, and are easy to leave accidentally in a loop. Never do something twice that only needs to be done once. That’s the worst kind of wasted time.

Keep Your Eye On APIs

What other external resources do you rely on? It’s a rare application that doesn’t leverage APIs. Unfortunately, in many or most cases, you don’t have control over the remote APIs you’re using, so you can’t do anything directly about their performance. You can, however, mitigate the effect of API performance in your code through techniques such as caching API output or making API calls in the background.

Your main goal is to protect your user from a failing or misbehaving API. Make sure you have reasonable timeouts in place for any API requests and, to the best of your application’s ability, be ready to display your application’s output without the API’s response.

Now Profile Your PHP

If you’re lucky, just enabling an opcode cache and optimizing external resource usage is enough to get the performance gains you need at the moment. But eventually, as your application needs increase, you’ll need or want to go deeper to get better performance to maintain or boost user experience and conserve hosting costs.

A number of open source tools can help you profile your PHP code and discover where the most time is being spent. One of the most common is Xdebug. Xdebug does require compiling and running special extensions on your server, and was originally intended, as its name implies, to be a debugging tool; the profiling aspects were added later.

It’s not easy to keep application performance in step with ever-increasing user expectations. But looking into and optimizing the functionality described above can help make sure your PHP application is performing at its best.

Gain better visibility and ensure optimal PHP application performance, try AppDynamics for FREE today!

Omed Habib
Omed Habib is a Director of Product Marketing at AppDynamics. He originally joined AppDynamics as a Principal Product Manager to lead the development of their world-class PHP, Node.js and Python APM agents. An engineer at heart, Omed fell in love with web-scale architecture while directing technology throughout his career. He spends his time exploring new ways to help some of the largest software deployments in the world meet their performance needs.

Thank you! Your submission has been received!

Oops! Something went wrong while submitting the form