Network tuning for Linux kernels
TCP/IP tuning is enabled by default for kernels after 2.6.17.
Check if auto-tuning is enabled in /proc/sys/net/ipv4/tcp_moderate_rcvbuf , it should be set to 1.
Also increase memory reserved for TCP send/receive buffers.
Initially “echo” the below values to the corresponding /proc file. If you see any considerable difference in the upload/download port speed, you can set it as a sysctl parameter.
net.ipv4.tcp_sack = 1
net.ipv4.tcp_timestamps = 0
net.ipv4.tcp_window_scaling = 1
net.core.rmem_max = 16777216
net.core.wmem_max = 16777216
net.ipv4.tcp_rmem = 4096 256960 16777216
net.ipv4.tcp_wmem = 4096 256960 16777216
net.ipv4.tcp_no_metrics_save = 1
net.core.rmem_default = 16777216
net.core.wmem_default = 16777216
net.ipv4.tcp_timestamps = 0
net.ipv4.tcp_window_scaling = 1
net.core.rmem_max = 16777216
net.core.wmem_max = 16777216
net.ipv4.tcp_rmem = 4096 256960 16777216
net.ipv4.tcp_wmem = 4096 256960 16777216
net.ipv4.tcp_no_metrics_save = 1
net.core.rmem_default = 16777216
net.core.wmem_default = 16777216
Leave a Reply
You must be logged in to post a comment.
