Boost Linux Network Performance: Proven Techniques to Increase Bandwidth and Reduce Latency

Key Metrics for Linux Network Tuning

Before tuning, understand the four critical metrics: bandwidth, latency, packet loss, and throughput.

Techniques to Increase Bandwidth

1. Adjust TCP window size

TCP window size controls the amount of data in flight. Increase the receive and send buffers to improve throughput.

sysctl net.ipv4.tcp_rmem
sysctl net.ipv4.tcp_wmem
sysctl -w net.ipv4.tcp_rmem="4096 87380 6291456"
sysctl -w net.ipv4.tcp_wmem="4096 65536 6291456"

2. Enable TCP Fast Open

TCP Fast Open reduces connection‑setup latency by allowing data to be sent during the three‑way handshake.

sysctl -w net.ipv4.tcp_fastopen=3

3. Change TCP congestion control algorithm

Check the current algorithm and switch to BBR for higher throughput on high‑bandwidth, high‑RTT links.

sysctl net.ipv4.tcp_congestion_control
sysctl -w net.ipv4.tcp_congestion_control=bbr

Techniques to Reduce Latency

1. Optimize kernel parameters

Increase buffer sizes and adjust timeout and reuse settings to shorten wait times.

sysctl -w net.ipv4.tcp_rmem="4096 87380 6291456"
sysctl -w net.ipv4.tcp_wmem="4096 65536 6291456"
sysctl -w net.ipv4.tcp_fin_timeout=30
sysctl -w net.ipv4.tcp_tw_reuse=1

2. Use efficient NIC drivers and irqbalance

Verify driver version and enable irqbalance to lower interrupt latency.

ethtool -i eth0

3. Zero‑copy transmission

Use the sendfile() system call to avoid copying data between user and kernel space.

sendfile(socket, file_descriptor, NULL, file_size);

4. High‑precision clock synchronization

Synchronize the server clock with NTP to eliminate time‑drift‑induced delays.

ntpq -p

Network Load Balancing and Optimization

1. Multi‑NIC bonding

Combine multiple network interfaces into a single logical bond to distribute traffic.

ifenslave bond0 eth0 eth1

2. Tune TCP connection limits

Raise the maximum number of pending connections to handle high concurrency.

sysctl -w net.core.somaxconn=65535
sysctl -w net.ipv4.tcp_max_syn_backlog=65535

Monitoring and Analysis Tools

iftop

/ nload – real‑time bandwidth usage per interface. netstat / ss – display active sockets and listening ports. iperf – measure bandwidth, latency, and packet loss between hosts.

Conclusion

By adjusting kernel parameters, selecting suitable TCP algorithms, employing zero‑copy, optimizing NICs, and using proper monitoring utilities, administrators can significantly increase bandwidth and lower latency, resulting in more stable and responsive services in data‑center or cloud environments.

Original source: https://blog.csdn.net/absurd/article/details/622055