Stop Letting Apache Eat Your RAM

If you are actively running a high-traffic Magento store or a Drupal installation on standard Apache mod_php right now, you are essentially setting money on fire. I recently audited a client's dedicated server here in Oslo. They were throwing hardware at a problem that was purely architectural. Their load average was hitting 20.0 on a quad-core box every time a crawler hit the site.

The culprit? The archaic process model of Apache's prefork MPM embedding the PHP interpreter into every single request.

It is May 2011. PHP 5.3.3 merged the FastCGI Process Manager (FPM) into the core. There is no excuse to ignore it anymore. If you want to survive the "Slashdot effect" or handle heavy traffic without your server swapping to disk, you need to decouple your web server from your PHP processing.

The Architecture: Nginx + PHP-FPM

Apache is great, but it is heavy. When serving static files (images, CSS, JS), an Apache child process still carries the baggage of the PHP interpreter if you use mod_php. That is 20MB to 50MB of RAM wasted per connection.

Enter Nginx paired with PHP-FPM. Nginx handles the connections (solving the C10k problem with an event-driven architecture) and passes only the PHP requests to the FPM socket. This keeps your memory footprint distinct and manageable.

The Configuration That Matters

Most tutorials give you generic settings. If you are hosting on a VPS with limited RAM (even a robust one), you need to tune the process manager. Do not use pm = static unless you know exactly how much RAM your application leaks.

Here is a battle-tested configuration for /etc/php-fpm.d/www.conf on a CentOS 5 or 6 box with 2GB RAM:

[www]
listen = /var/run/php-fpm/php-fpm.sock
listen.owner = nginx
listen.group = nginx
user = nginx
group = nginx

; The process manager control
pm = dynamic
pm.max_children = 50
pm.start_servers = 5
pm.min_spare_servers = 5
pm.max_spare_servers = 35

; Kill processes that leak memory after 500 requests
pm.max_requests = 500

; Slow log is critical for debugging bad code
request_slowlog_timeout = 5s
slowlog = /var/log/php-fpm/www-slow.log

Pro Tip: Always use Unix sockets (.sock) instead of TCP ports (127.0.0.1:9000) if the web server and PHP are on the same machine. The TCP overhead adds unnecessary latency. In a test environment connected to NIX (Norwegian Internet Exchange), we saw a 10% throughput drop using TCP loopback versus sockets.

Storage I/O: The Silent Killer

You can tune PHP-FPM all day, but if your disk I/O is trash, your site will hang. PHP sessions and MySQL temporary tables thrash the disk.

In 2011, relying on standard SATA 7200RPM drives for a database-heavy application is negligence. We are seeing a massive shift toward Solid State Drives (SSD) in the enterprise sector. While expensive, the IOPs (Input/Output Operations Per Second) are vastly superior.

This is where infrastructure choice dictates success. At CoolVDS, we have started implementing high-performance SSD storage arrays. While others pack you onto crowded nodes with slow SATA drives, our architecture is built to prevent "noisy neighbors" from stealing your I/O cycles. When PHP-FPM writes a session file, it happens instantly. No wait states.

Data Integrity and The Law

Performance isn't just about speed; it's about control. With the Personopplysningsloven (Personal Data Act) and the watchful eye of Datatilsynet, you cannot afford to have data leaking or servers crashing.

Using PHP-FPM allows you to chroot individual pools. This means if one site on your server gets compromised via an SQL injection, the attacker is trapped in that specific directory. They cannot traverse up to /etc/ and read your system configs. This level of isolation is standard practice for any serious sysadmin in Norway handling sensitive customer data.

The Migration Path

1. Install Nginx: Add the EPEL repository if you are on CentOS.
2. Install PHP-FPM: yum install php-fpm. Ensure you are on PHP 5.3+.
3. Configure Nginx: Point your fastcgi_pass to the unix socket defined above.
4. Test: Use ab (Apache Bench) to hammer the site. Watch top. Smile as your load average stays below 1.0.

Do not let legacy configurations dictate your uptime. If you need a sandbox to test this setup without risking your production hardware, spin up a CoolVDS instance. We offer clean Linux templates ready for serious engineering.

Get your load times under control. Deploy a high-performance SSD VPS with CoolVDS today.