How to Enable Google BBR TCP Congestion Control on Ubuntu VPS for 30% Faster Network Speeds

The Problem with Traditional TCP CUBIC

By default, almost every modern Linux distribution (Ubuntu 22.04, 24.04, Debian 12, CentOS/AlmaLinux) ships with CUBIC as its default TCP congestion control algorithm. CUBIC was designed in an era of loss-based congestion detection: it assumes that whenever a network packet is dropped, the network is congested, causing it to slash its transmission window (speed) by up to 50%.

On modern long-distance internet connections—such as a user in Singapore accessing a VPS hosted in Frankfurt, Germany—minor wireless interference or momentary routing jitter causes harmless packet loss. CUBIC panics and throttles throughput, leaving massive bandwidth capacity completely unused.

What is Google BBR?

BBR (Bottleneck Bandwidth and RTT) is an algorithm created by Google network engineers that powers YouTube, Google Search, and Google Cloud. Instead of treating packet loss as congestion, BBR builds an active mathematical model of the physical network pipe: it measures maximum available bottleneck bandwidth and minimum round-trip time (RTT). It pushes data at the exact physical capacity of the link without overflowing network buffers.

Step 1: Check Current Kernel & Congestion Algorithm

Google BBR requires Linux Kernel version 4.9 or newer. Modern Ubuntu VPS instances run Linux 6.x+, well above the requirement. Verify with:

uname -r
# Output: 6.8.0-31-generic (or similar)

Check your currently active TCP congestion algorithm:

sysctl net.ipv4.tcp_congestion_control
# Output: net.ipv4.tcp_congestion_control = cubic

List all available algorithms supported by your current kernel:

sysctl net.ipv4.tcp_available_congestion_control
# Output: net.ipv4.tcp_available_congestion_control = reno cubic bbr

Step 2: Activating TCP BBR via sysctl.conf

To enable BBR permanently across server reboots, append the Fair Queueing (FQ) packet scheduler and BBR congestion control module to /etc/sysctl.conf:

cat << 'EOF' >> /etc/sysctl.conf

# Google BBR TCP Congestion Control & Buffer Tuning
net.core.default_qdisc = fq
net.ipv4.tcp_congestion_control = bbr
net.core.rmem_max = 16777216
net.core.wmem_max = 16777216
net.ipv4.tcp_rmem = 4096 87380 16777216
net.ipv4.tcp_wmem = 4096 65536 16777216
EOF

Step 3: Applying Kernel Settings

Apply the configuration immediately without rebooting the server:

sysctl -p

Step 4: Verifying BBR is Active and Running

Confirm the new congestion controller is active:

sysctl net.ipv4.tcp_congestion_control
# Expected output: net.ipv4.tcp_congestion_control = bbr

Verify that the kernel module is loaded in memory:

lsmod | grep bbr
# Output: tcp_bbr                20480  24

If tcp_bbr appears in the module list, Google BBR is actively handling all inbound and outbound TCP network sockets on your server.

Benchmarking with iperf3

To witness the throughput disparity firsthand, spin up an iperf3 test between your local machine and your VPS under synthetic 1% packet loss conditions. While CUBIC collapses to ~3.5 Mbps, BBR sustains over 120+ Mbps by refusing to artificially throttle when the physical connection pipe is clear.

Conclusion

Activating Google BBR is the highest return-on-investment optimization available for any Linux server. Whether you are hosting a WordPress site on Hostinger KVM VPS or handling API traffic on Contabo Cloud VPS, taking 90 seconds to activate BBR transforms network responsiveness globally.