Linux VPS Performance Tuning: Optimize Kernel Parameters for Maximum Throughput

If you’re running applications on a Linux VPS, you’re already getting dedicated resources and root access. But are you getting the maximum performance your virtual server can deliver? Default Linux kernel settings are tuned for general-purpose use, not for the specific workloads you’re running. By adjusting key kernel parameters, you can significantly improve network throughput, disk I/O, and memory management on your VPS. For a comparison of VPS providers that support full kernel tuning, check out the provider comparison table.

Understanding sysctl and Kernel Parameters

The Linux kernel exposes hundreds of tunable parameters through the /proc/sys/ virtual filesystem. You can view and modify them at runtime using the sysctl command. Permanent changes go into /etc/sysctl.conf or a file in /etc/sysctl.d/. Before making any changes, always back up your current configuration:

sudo sysctl -a > ~/sysctl-backup-$(date +%F).txt

1. Network Performance Tuning

Network latency and throughput are critical for web servers, APIs, and database-driven applications. Here are the most impactful tweaks:

Enable TCP BBR Congestion Control

BBR (Bottleneck Bandwidth and Round-trip propagation time) is Google’s modern congestion control algorithm. It can dramatically improve throughput on high-latency links:

echo 'net.core.default_qdisc=fq' | sudo tee -a /etc/sysctl.conf
echo 'net.ipv4.tcp_congestion_control=bbr' | sudo tee -a /etc/sysctl.conf
sudo sysctl -p

Verify BBR is active with sysctl net.ipv4.tcp_congestion_control. You should see bbr as the output.

Increase TCP Buffer Sizes

Default TCP buffer sizes are conservative. For better throughput, especially on connections with high bandwidth-delay product, increase them:

# Increase TCP max buffer sizes
net.core.rmem_max = 134217728
net.core.wmem_max = 134217728
# Set TCP auto-tuning buffers (min, default, max)
net.ipv4.tcp_rmem = 4096 87380 134217728
net.ipv4.tcp_wmem = 4096 65536 134217728
# Enable TCP window scaling
net.ipv4.tcp_window_scaling = 1

Optimize Connection Handling

For web servers handling many concurrent connections, these settings prevent socket exhaustion:

net.ipv4.tcp_fin_timeout = 15
net.ipv4.tcp_tw_reuse = 1
net.core.somaxconn = 4096
net.ipv4.tcp_max_syn_backlog = 8192
net.core.netdev_max_backlog = 5000

2. Memory and VM Tuning

VPS instances often have limited RAM, so memory management settings matter significantly:

# Reduce swappiness (default 60) — prefer keeping data in RAM
vm.swappiness = 10
# Increase dirty page ratios for better write performance
vm.dirty_background_ratio = 5
vm.dirty_ratio = 10
# Increase max number of memory map areas a process may have
vm.max_map_count = 262144

A swappiness value of 10 tells the kernel to avoid swapping unless absolutely necessary. This is particularly useful on VPS plans with limited disk I/O, where swap thrashing can cripple performance. Compare VPS plans with sufficient RAM for your workload at the provider comparison page.

3. Filesystem and Disk I/O Tuning

Disk I/O is often the bottleneck on virtual servers. These settings help optimize filesystem performance:

# Increase the maximum number of open files
fs.file-max = 2097152
# Allow more inotify watchers (useful for file synchronization tools)
fs.inotify.max_user_watches = 524288
# Increase aio (async I/O) request limit
fs.aio-max-nr = 1048576

Choosing the Right I/O Scheduler

Modern Linux kernels use the multi-queue block layer. For VPS instances backed by SSD storage (common with KVM-based providers), use the none or nvme scheduler. For traditional spinning disks, kyber or bfq may work better. Check your current scheduler with:

cat /sys/block/*/queue/scheduler

To switch to the none scheduler (best for SSDs):

echo 'ACTION=="add|change", KERNEL=="sd*[!0-9]", ATTR{queue/scheduler}="none"' | sudo tee /etc/udev/rules.d/60-iosched.rules

4. Applying Changes Permanently

Create a dedicated configuration file and apply all settings at once:

sudo tee /etc/sysctl.d/99-vps-performance.conf << 'EOF'
# Network tuning
net.core.default_qdisc = fq
net.ipv4.tcp_congestion_control = bbr
net.core.rmem_max = 134217728
net.core.wmem_max = 134217728
net.ipv4.tcp_rmem = 4096 87380 134217728
net.ipv4.tcp_wmem = 4096 65536 134217728
net.ipv4.tcp_window_scaling = 1
net.ipv4.tcp_fin_timeout = 15
net.ipv4.tcp_tw_reuse = 1
net.core.somaxconn = 4096
net.ipv4.tcp_max_syn_backlog = 8192
net.core.netdev_max_backlog = 5000
# Memory tuning
vm.swappiness = 10
vm.dirty_background_ratio = 5
vm.dirty_ratio = 10
vm.max_map_count = 262144
# Filesystem tuning
fs.file-max = 2097152
fs.inotify.max_user_watches = 524288
fs.aio-max-nr = 1048576
EOF
sudo sysctl -p /etc/sysctl.d/99-vps-performance.conf

5. Verifying Performance Gains

After applying these optimizations, benchmark your VPS to measure improvement:

• Network: Use iperf3 or speedtest-cli to test throughput before and after.
• Disk I/O: Run fio with sequential and random read/write tests.
• Latency: Use ping and mtr to check round-trip times to key endpoints.

These kernel-level optimizations are just the start. For maximum VPS performance, pair them with application-level tuning (database query optimization, caching layers, CDN integration) and choose a VPS provider with modern hardware. Browse VPS provider comparisons to find a host that gives you full kernel control and high-performance SSD storage.