Dodaj do ulubionych

optymalizacja linux TCP stack

15.02.05, 17:56
Po raz pierwszy od dawna zainstalowalem nowego linuksa: SuSe 9.2 64-bit.
Z przyzwyczajenia zaczalem robic standardowe optymalizacje. W
efekcie SuSe 9.2 z KDE po starcie zajmuje 130MB RAM, kernel laduje sie ~45sec.
Ustawilem vm.swappiness na 0 (przy 1GB RAM i przeznaczeniu na workstation a
nie serwer) itd itp. w sumie jestem dosc zadowolony (po ~dwoch tygodniach
uzywania :))

Problem. w testach szybkosciowych sieci uzyskuje srednio na windows XP
375KB/sec (czyli max dopuczszalne przez mojego ISP provider'a) ale tylko
300KB/sec pod SuSe.
sprzet:
windows XP
AMD XP2800, 512MB RAM, NIC Linksys LNE100, HDD WD ATA 80GB
SuSe 64-bit
AMD64 3000, 1GB RAM, NIC Linksys LNE100, HDD Hitachi SATA 120GB
na obu FF z takimi samymi ustawieniami.
na obu
Window scaling off
sack on
PMTU on
Timestamping off
MTU 1500

Ciekawi mnie co jeszcze moge zmienic aby poprawic wydajnosc sieci pod linuksem?


szczegolowe ustawienia SuSe wygladaja tak:
# Turn off the tcp_window_scaling if you have networking
# with systems behind broken routers, otherwise keep it on (default)
net.ipv4.tcp_window_scaling = 0
#Enable Path MTU Discovery. On by default
#net.ipv4.ip_no_pmtu_disc = 0
# Leave the tcp_sack on (default)
#net.ipv4.tcp_sack = 1
# tcp_fack should be changed too because of sack. Default "on"
#net.ipv4.tcp_fack = 1
# Turn off the tcp_timestamps
net.ipv4.tcp_timestamps = 0
# Set TCP Re-Ordering value in kernel to 5
net.ipv4.tcp_reordering = 5
#Decrease SYN ACK retry attempts to 2
net.ipv4.tcp_synack_retries = 2
net.ipv4.tcp_syn_retries = 3
# Decrease the time default value for tcp_fin_timeout connection (*)
net.ipv4.tcp_fin_timeout = 25
# Increase the tcp-time-wait buckets pool size (*)
net.ipv4.tcp_max_tw_buckets = 360000
# Enable TCP SYN Cookie Protection (*)
net.ipv4.tcp_syncookies = 1
net.ipv4.ip_forward = 0 (*)
# Set Max SYN Backlog (*)
net.ipv4.tcp_max_syn_backlog = 1024
net.ipv4.core.netdev_max_backlog = 1024
# This will increase the amount of memory available for socket input/output
queues.
#These numbers can be adjusted depending of the machine
net.core.optmem_max = 40960
net.ipv4.core.rmem_max = 8388608
net.ipv4.core.rmem_default = 65535
net.ipv4.core.wmem_max = 8388608
net.ipv4.core.wmem_default = 65535
net.ipv4.tcp_rmem = 4096 87380 8388608
net.ipv4.tcp_wmem = 4096 87380 8388608
net.ipv4.tcp_mem = 8388608 8388608 8388608

## Controls source route verification (*)
net.ipv4.conf.default.rp_filter = 1

# Enable bad error message Protection (*)
net.ipv4.icmp_ignore_bogus_error_responses = 1

# Disable ICMP broadcasts request (*)
net.ipv4.icmp_echo_ignore_broadcasts = 1
# Disables IP source routing (*)
net.ipv4.conf.all.default_source_route = 0
net.ipv4.conf.default.accept_redirects = 0
net.ipv4.conf.default.secure_redirects = 0
net.ipv4.conf.all.secure_redirects = 0
#Enable log of spoofed packets
net.ipv4.conf.all.log_martians = 1
# Decrease time between keepalives (*)
net.ipv4.tcp_keepalive_time = 1200
# Controls the System Request debugging functionality of the kernel
kernel.sysrq = 0
Obserwuj wątek

Nie pamiętasz hasła

lub ?

 

Nie masz jeszcze konta? Zarejestruj się

Nakarm Pajacyka