BBR (Bottleneck Bandwidth and Round-trip propagation time) is Google’s TCP congestion control algorithm that significantly improves throughput and reduces latency compared to default Linux algorithms like CUBIC. On long-distance or high-latency connections, BBR can deliver 2-3x higher throughput<\/strong> and 40-60% lower latency<\/strong>. This guide walks through enabling BBR on any modern Linux VPS.<\/p>\n\n\n\n Traditional congestion control (CUBIC, Reno) detects congestion only after<\/em> packet loss occurs \u2014 they keep increasing send rate until the buffer fills and drops packets. BBR models<\/em> the network path by measuring bandwidth and RTT in real time, then paces packets at the exact rate the link can sustain. The result: no bufferbloat, lower retransmit rates, and consistently higher throughput on lossy or high-latency paths.<\/p>\n\n\n\n First, check what algorithm your VPS currently uses:<\/p>\n\n\n\n On most modern kernels (4.9+), BBR is either built-in or available as a loadable module. If Check your kernel version:<\/strong> Edit sysctl configuration:<\/p>\n\n\n\n Enterprise Linux 8+ kernels (4.18+) include BBR support natively:<\/p>\n\n\n\n After applying, confirm the change took effect:<\/p>\n\n\n\n Run a quick benchmark using For real-world testing, use What BBR Does Differently<\/h2>\n\n\n\n
Checking Your Current Congestion Control<\/h2>\n\n\n\n
# Check available congestion control algorithms\nsysctl net.ipv4.tcp_available_congestion_control\n\n# Check current active algorithm\nsysctl net.ipv4.tcp_congestion_control\n\n# Check if BBR module is available\nlsmod | grep tcp_bbr\n<\/code><\/pre>\n\n\n\nlsmod | grep tcp_bbr<\/code> returns nothing, you can load it manually.<\/p>\n\n\n\nStep 1: Enable BBR on Ubuntu \/ Debian<\/h2>\n\n\n\n
uname -r<\/code> \u2014 BBR requires kernel 4.9 or newer. Ubuntu 18.04+ and Debian 10+ satisfy this.<\/p>\n\n\n\n# Load BBR module\nmodprobe tcp_bbr\n\n# Make it persistent\necho \"tcp_bbr\" >> \/etc\/modules-load.d\/bbr.conf\n\n# Set BBR as the active congestion control\necho \"net.core.default_qdisc = fq\" >> \/etc\/sysctl.d\/bbr.conf\necho \"net.ipv4.tcp_congestion_control = bbr\" >> \/etc\/sysctl.d\/bbr.conf\n\n# Apply\nsysctl -p \/etc\/sysctl.d\/bbr.conf\n<\/code><\/pre>\n\n\n\nStep 2: Enable BBR on CentOS \/ Rocky \/ AlmaLinux<\/h2>\n\n\n\n
# Load module\nmodprobe tcp_bbr\n\n# Persistent config\necho \"tcp_bbr\" | tee \/etc\/modules-load.d\/99-bbr.conf\n\n# Apply via sysctl\nsysctl -w net.core.default_qdisc=fq\nsysctl -w net.ipv4.tcp_congestion_control=bbr\n\n# Make persistent\ncat >> \/etc\/sysctl.d\/99-bbr.conf <<EOF\nnet.core.default_qdisc = fq\nnet.ipv4.tcp_congestion_control = bbr\nEOF\n<\/code><\/pre>\n\n\n\nStep 3: Verify BBR Is Active<\/h2>\n\n\n\n
sysctl net.ipv4.tcp_congestion_control\n# Output should show: net.ipv4.tcp_congestion_control = bbr\n\n# Also check loaded modules\nlsmod | grep bbr\n# Output should show: tcp_bbr\n<\/code><\/pre>\n\n\n\nBenchmarking the Improvement<\/h2>\n\n\n\n
iperf3<\/code> to compare before and after:<\/p>\n\n\n\n# On a remote server (or use a public iperf server):\niperf3 -c iperf.he.net -t 30\n\n# BBR typically shows 2-3x the throughput of CUBIC\n# on paths with 50ms+ RTT or any packet loss\n<\/code><\/pre>\n\n\n\ncurl -o \/dev\/null -s -w '%{speed_download}\n' https:\/\/testfile.example.com<\/code> to measure download speeds from your actual content sources.<\/p>\n\n\n\nAdditional Tuning for Maximum BBR Performance<\/h2>\n\n\n\n