Linux automatically caches frequently accessed files in unused RAM through its page cache system, but we can take more control over this process. The kernel's virtual memory subsystem maintains this cache transparently, serving subsequent requests from RAM instead of disk.

# Check current cache usage
free -h
# Output example:
#               total        used        free      shared  buff/cache   available
# Mem:           3.9G        1.1G        500M         50M        2.3G        2.4G
# Swap:          2.0G         50M        2.0G

Using vmtouch - The Virtual Memory Toucher

vmtouch is a perfect tool for portable cache control:

# Install vmtouch on Debian/Ubuntu
sudo apt-get install vmtouch

# Load entire directory into cache
vmtouch -t /var/www/html/

# Check what's currently cached
vmtouch /var/www/html/main.css
# Output: /var/www/html/main.css is 92% cached

Direct POSIX Methods

For programmatic control, use POSIX_FADV_WILLNEED:

#include <fcntl.h>
#include <unistd.h>

void preload_file(const char *path) {
    int fd = open(path, O_RDONLY);
    posix_fadvise(fd, 0, 0, POSIX_FADV_WILLNEED);
    close(fd);
}

Tune swappiness and cache pressure for better results:

# Reduce swap tendency (0-100, lower values avoid swapping)
echo 10 > /proc/sys/vm/swappiness

# Increase cache retention (0-100, higher values keep cache longer)
echo 50 > /proc/sys/vm/vfs_cache_pressure

# Make changes permanent
echo "vm.swappiness = 10" >> /etc/sysctl.conf
echo "vm.vfs_cache_pressure = 50" >> /etc/sysctl.conf

Create a service to cache files at boot:

# /etc/systemd/system/precache.service
[Unit]
Description=Preload important files into RAM cache
After=network.target

[Service]
Type=oneshot
ExecStart=/usr/bin/vmtouch -dl /var/www/html/ /usr/lib/large_library/

[Install]
WantedBy=multi-user.target

Use tools to verify your cache strategy:

# Install and use cachestat
sudo apt-get install perf-tools-unstable
sudo cachestat 1  # 1-second intervals

# Alternative using /proc
grep -E '^(Dirty|Writeback):' /proc/meminfo

For special cases consider:

• tmpfs: mount -t tmpfs -o size=1G tmpfs /mnt/ramdisk
• ramfs: mount -t ramfs ramfs /mnt/ramdisk (no size limits)
• mlock(): System call to prevent paging of critical files

Linux automatically caches frequently accessed files in RAM through its page cache mechanism. However, we can optimize this behavior for specific use cases. Here's how to actively control what gets cached:

# Check current cache usage
free -h
cat /proc/meminfo | grep -i cache

Using vmtouch - The Virtual Memory Toucher

This excellent tool gives you precise control over file caching:

# Install vmtouch on Debian/Ubuntu
sudo apt-get install vmtouch

# Load entire directory into cache
vmtouch -t /var/www/html/

# Lock files in RAM (prevent eviction)
vmtouch -l /var/www/html/important_files/

Kernel-Level Control with fincore

For more advanced control, examine and manipulate the page cache directly:

# Check which pages are cached
fincore /var/www/html/*.js

# Preload files using dd
dd if=/path/to/file of=/dev/null bs=1M

Systemd Service for Automatic Preloading

Create a service to cache files at boot:

# /etc/systemd/system/preload.service
[Unit]
Description=Preload web files into RAM
After=network.target

[Service]
Type=oneshot
ExecStart=/usr/bin/vmtouch -dl /var/www/html/

[Install]
WantedBy=multi-user.target

Using preload Daemon

The preload package predicts and caches frequently used applications:

sudo apt install preload
sudo systemctl enable --now preload

RAM Disk with Transparent Fallback

For mission-critical files, consider tmpfs with automatic HDD fallback:

# /etc/fstab entry
tmpfs /var/www/html/critical tmpfs defaults,size=2G 0 0

# Mirror files with rsync
rsync -a --delete /var/www/html/critical/ /var/www/html-copy/

Kernel Parameter Tuning /h2>

Optimize cache behavior via sysctl:

# Increase cache pressure (lower = keep more cached)
vm.vfs_cache_pressure=50

# Reduce swap tendency (avoid caching to disk)
vm.swappiness=10

Essential commands to verify cache effectiveness:

# Check cached files
vmtouch -v /var/www/html/

# Monitor cache hits
sudo perf stat -e cache-references,cache-misses -a sleep 5

# Detailed page cache stats
cat /proc/meminfo | grep -i cached