主 题：发一个美西（或者欧洲）高延时bbr和tcp优化方案
发 布 者：rosnode
标签分类： 技术
时 间：2026-07-13 14:47:12
内容预览：开启并优化 BBR v1 + fqnet.core.default_qdisc = fqnet.ipv4.tcp_congestion_control = bbr大缓存与队列优化 (适合美西长肥管道)net.core.rmem_max = 67108864net.core.wmem_max = 67108864net.core.rmem_default = 33554432net.core.wmem_default = 33554432net.ipv4.tcp_rmem = 4096 87380 67108864net.ipv4.tcp_wmem = 4096 65536 67108864net.core.netdev_max_backlog = 10000net.core.somaxconn = 4096减少高延迟握手消耗net.ipv4.tcp_fastopen = 3net.ipv4.tcp_window_scaling = 1丢包与重传控制net.ipv4.tcp_sack = 1net.ipv4.tcp_dsack = 1net.ipv4.tcp_tw_reuse = 1net.ipv4.tcp_fin_timeout = 15net.ipv4.tcp_retries2 = 8别用v3 我仔细对比过单线程测速，优化过以后 可以得到大幅度提升。
直达链接： https://www.nodeseek.com/post-819757-1
发 布 者：rosnode
标签分类： 技术
时 间：2026-07-13 14:47:12
内容预览：开启并优化 BBR v1 + fqnet.core.default_qdisc = fqnet.ipv4.tcp_congestion_control = bbr大缓存与队列优化 (适合美西长肥管道)net.core.rmem_max = 67108864net.core.wmem_max = 67108864net.core.rmem_default = 33554432net.core.wmem_default = 33554432net.ipv4.tcp_rmem = 4096 87380 67108864net.ipv4.tcp_wmem = 4096 65536 67108864net.core.netdev_max_backlog = 10000net.core.somaxconn = 4096减少高延迟握手消耗net.ipv4.tcp_fastopen = 3net.ipv4.tcp_window_scaling = 1丢包与重传控制net.ipv4.tcp_sack = 1net.ipv4.tcp_dsack = 1net.ipv4.tcp_tw_reuse = 1net.ipv4.tcp_fin_timeout = 15net.ipv4.tcp_retries2 = 8别用v3 我仔细对比过单线程测速，优化过以后 可以得到大幅度提升。
直达链接： https://www.nodeseek.com/post-819757-1
