Tuning for Better Network Performance

Discussion in 'all things UNIX' started by Searching_ _ _, Nov 13, 2011.

Thread Status:
Not open for further replies.
  1. Searching_ _ _

    Searching_ _ _ Registered Member

    Joined:
    Jan 2, 2008
    Posts:
    1,988
    Location:
    iAnywhere
    I'm interested in finding out if I can improve on my existing network performance.
    Can turning off the 12 bytes of timestamp really be a meaningful change for performance?
    How do you tweak the RWIN in Linux exactly?

    I've looked at these sites but don't feel comfortable without further advice.
    http://www.psc.edu/networking/projects/tcptune/
    I was unable to follow along with: http://www.linuxforums.org/forum/networking/124571-solved-changing-rwin-value.html

    These are some suggestions:
    These are some values inside the system:

    Code:
    $ sysctl -a | fgrep tcp
    error: permission denied on key 'kernel.cad_pid'
    error: permission denied on key 'fs.binfmt_misc.register'
    error: permission denied on key 'dev.parport.parport0.autoprobe'
    error: permission denied on key 'dev.parport.parport0.autoprobe0'
    error: permission denied on key 'dev.parport.parport0.autoprobe1'
    error: permission denied on key 'dev.parport.parport0.autoprobe2'
    error: permission denied on key 'dev.parport.parport0.autoprobe3'
    error: permission denied on key 'net.ipv4.route.flush'
    net.ipv4.tcp_timestamps = 1
    net.ipv4.tcp_window_scaling = 1
    net.ipv4.tcp_sack = 1
    net.ipv4.tcp_retrans_collapse = 1
    net.ipv4.tcp_syn_retries = 5
    net.ipv4.tcp_synack_retries = 5
    net.ipv4.tcp_max_orphans = 32768
    net.ipv4.tcp_max_tw_buckets = 180000
    net.ipv4.tcp_keepalive_time = 7200
    net.ipv4.tcp_keepalive_probes = 9
    net.ipv4.tcp_keepalive_intvl = 75
    net.ipv4.tcp_retries1 = 3
    net.ipv4.tcp_retries2 = 15
    net.ipv4.tcp_fin_timeout = 60
    net.ipv4.tcp_syncookies = 1
    net.ipv4.tcp_tw_recycle = 0
    net.ipv4.tcp_abort_on_overflow = 0
    net.ipv4.tcp_stdurg = 0
    net.ipv4.tcp_rfc1337 = 0
    net.ipv4.tcp_max_syn_backlog = 1024
    net.ipv4.tcp_orphan_retries = 0
    net.ipv4.tcp_fack = 1
    net.ipv4.tcp_reordering = 3
    net.ipv4.tcp_ecn = 2
    net.ipv4.tcp_dsack = 1
    net.ipv4.tcp_mem = 80832    107776    161664
    net.ipv4.tcp_wmem = 4096    16384    3448832
    net.ipv4.tcp_rmem = 4096    87380    3448832
    net.ipv4.tcp_app_win = 31
    net.ipv4.tcp_adv_win_scale = 2
    net.ipv4.tcp_tw_reuse = 0
    net.ipv4.tcp_frto = 2
    net.ipv4.tcp_frto_response = 0
    net.ipv4.tcp_low_latency = 0
    net.ipv4.tcp_no_metrics_save = 0
    net.ipv4.tcp_moderate_rcvbuf = 1
    net.ipv4.tcp_tso_win_divisor = 3
    net.ipv4.tcp_congestion_control = cubic
    net.ipv4.tcp_abc = 0
    net.ipv4.tcp_mtu_probing = 0
    net.ipv4.tcp_base_mss = 512
    net.ipv4.tcp_workaround_signed_windows = 0
    net.ipv4.tcp_dma_copybreak = 4096
    net.ipv4.tcp_slow_start_after_idle = 1
    net.ipv4.tcp_available_congestion_control = cubic reno
    net.ipv4.tcp_allowed_congestion_control = cubic reno
    net.ipv4.tcp_max_ssthresh = 0
    net.ipv4.tcp_cookie_size = 0
    net.ipv4.tcp_thin_linear_timeouts = 0
    net.ipv4.tcp_thin_dupack = 0
    error: permission denied on key 'net.ipv6.route.flush'

     
  2. Mrkvonic

    Mrkvonic Linux Systems Expert

    Joined:
    May 9, 2005
    Posts:
    8,698
    Don't go there. If you really don't know the impact of all those, the chances are fairly high you will screw things. And the best way to improve your network performance is a good isp and decent bandwidth.
    Mrk
     
  3. Searching_ _ _

    Searching_ _ _ Registered Member

    Joined:
    Jan 2, 2008
    Posts:
    1,988
    Location:
    iAnywhere
    Thanks for the warning.
    I've already changed these settings and have achieved 150+kb/s improvement. It's not permanent so all will revert when I reboot.
    How do I optamize? Should all four listed below be the same value? Would different values be better?
    Cheers
     
    Last edited: Nov 13, 2011
Loading...
Thread Status:
Not open for further replies.