Performance Tuning & Kernel Parameters (Sysctl)

Reducing TCP Buffer Bloat via Autotuning Limits

🧩 The Challenge

High-throughput network applications on modern 10Gbps+ links often experience performance bottlenecks due to static TCP buffer defaults being too small to handle long-fat network pipes. Manual adjustment of individual buffers is inefficient and risks system instability under variable memory loads.

💡 The Fix

Increase the kernel TCP autotuning memory limits to allow the networking stack to dynamically allocate larger buffers during high-bandwidth transmission, improving overall throughput.

sysctl -w net.ipv4.tcp_rmem="4096 87380 16777216"
sysctl -w net.ipv4.tcp_wmem="4096 65536 16777216"
sysctl -w net.core.rmem_max=16777216
sysctl -w net.core.wmem_max=16777216

⚙️ Why It Works

These parameters define the minimum, default, and maximum memory bytes allowed for TCP receive and send buffers, enabling the kernel to scale window sizes appropriately for faster network transfers. The new maximums allow for significantly larger TCP windows that prevent the sender from idling while waiting for acknowledgments.

🚀 Pro-Tip: Persist these settings across reboots by adding them to /etc/sysctl.d/99-network-performance.conf and running sysctl -p.

Linux Tips & Tricks | © ngelinux.com | 7/10/2026