There is any recommandation to improve the tcp\ip stack in redhat5.1, by tune any of it's parameters?

Thanks

If you mean "Red Hat Linux" (RHL) then you're SOL and you should upgrade ASAP. If you mean "Red Hat Enterprise Linux" (RHEL) then next time please use the right description. Could you post compelling reasons *why* you think you need to fiddle with a network stack that does a lot of autotuning?

I have RHEL5.1

I am porting solaris10 server to Linux RHEL5.1.
In my solaris10 server we configured some tcp\ip for tuning, like :
ndd -set /dev/tcp tcp_ip_abort_interval 20000
ndd -set /dev/tcp tcp_time_wait_interval 10000
ndd -set /dev/tcp tcp_rexmit_interval_max 5000
ndd -set /dev/tcp tcp_rexmit_interval_initial 2000
ndd -set /dev/tcp tcp_fin_wait_2_flush_interval 20000
ndd -set /dev/tcp tcp_conn_req_max_q0 65536
ndd -set /dev/tcp tcp_conn_req_max_q 65536
ndd -set /dev/tcp tcp_time_wait_interval 15000
ndd -set /dev/arp arp_cleanup_interval 60000

So I am trying to understand if there are some recommended tunes for the tcp\ip stack in RHEL5.1, in order to do right porting.

thanks

I'm sorry but I don't want to wade through the all of the soltuneparamref to find an explanation. These stacks AFAIK differ anyway so I doubt you'll find perfectly corresponding values. I'll redirect you to the ipsysctl tutorial instead where you can find the explanation for kernel IP stack tunables. Also if you use your favourite searchengine with terms like "Linux ip sysctl" or "linux network perf" or "Linux GigE perf" or "Linux congestion control" (Westwood, Vegas, BIC) you'll find some settings for tunables as well (also see: http://ipsysctl-tutorial.frozentux.n...TCPTUNINGGUIDE, http://www-didc.lbl.gov/TCP-tuning/linux.html, http://www.psc.edu/networking/perf_tune.html, http://www.onlamp.com/lpt/a/6324) and prolly searching LQ will turn up some as well. As I said before Linux does autotune a lot and a lot of these sysctls are related so be sure you understand what to tweak because it's easy to decrease performance.
Do some tests so you can check your tweaks against baseline results.

thank for the links