Date: Mon, 11 Dec 1995 16:41:22 +0200 From: Petri Helenius <pete@sms.fi> To: Chris Stenton <jacs@gnome.co.uk> Cc: questions@FreeBSD.ORG Subject: Why is my FreeBSD box so slow on local TCP? Message-ID: <199512111441.QAA14944@silver.sms.fi> In-Reply-To: <jacs-9511111231.AA00012651@hawk.gnome.co.uk> References: <jacs-9511111231.AA00012651@hawk.gnome.co.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
Chris Stenton writes: > > > I posted this on usenet but got no reply anyone care to give an answer? > > > I have just run lmbench on my pentium FreeBSD-stable box. > > Ok, I don't expect my box to be as fast as an HP snake but I did not > expect it to be 24 times slower than Linux on Local TCP connections. > Anyone any ideas whats going on, 100 kb/s is not exactly pushing the > boat out? > What's your lo0 MTU? If it's the 16384 that some non-tcp-knowledgeable person put in sometime in the past I think what you are seeing is called "TCP deadlock" which appears when window size is equal or smaller than the MTU. This makes TCP to be stop-and-go protocol (remember XMODEM? or non-windowing kermit) and thus the troughput of the protocol is pretty horrible. This happens on ATM also if you are running with 4096 or 8192 window sizes and the RFC1577 default MTU of 9180. Fortunately not too many applications use the 127.0.0.1 address but use the loopback provided by the ethernet-interface. (and thus get the MTU of 1500) Pete
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199512111441.QAA14944>