Return to BSD News archive
Newsgroups: comp.os.linux.advocacy,comp.unix.bsd.freebsd.misc,comp.unix.advocacy,comp.unix.misc,comp.unix.pc-clone.32bit Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!lll-winken.llnl.gov!uwm.edu!cs.utexas.edu!news.sprintlink.net!news1!not-for-mail From: root@dyson.iquest.net (John S. Dyson) Subject: Re: Linux vs FreeBSD X-Nntp-Posting-Host: dyson.iquest.net Message-ID: <4949f5$25t@dyson.iquest.net> Sender: news@iquest.net (News Admin) Organization: John S. Dyson's home machine References: <489kuu$rbo@pelican.cs.ucla.edu> <48ajsj$f5p@galaxy.ucr.edu> <490p3m$6c7@news.iii.net> <30B47F81.41C67EA6@FreeBSD.org> Date: Fri, 24 Nov 1995 11:14:45 GMT Lines: 37 Xref: euryale.cc.adfa.oz.au comp.os.linux.advocacy:28690 comp.unix.bsd.freebsd.misc:9570 comp.unix.advocacy:11533 comp.unix.misc:19722 comp.unix.pc-clone.32bit:8900 In article <30B47F81.41C67EA6@FreeBSD.org>, Jordan K. Hubbard <jkh@FreeBSD.org> wrote: >Craig Shrimpton wrote: >> >> In article <48ajsj$f5p@galaxy.ucr.edu>, >> >> > >> >They are both bitchen. >> > >> I think FreeBSD excels in fast networking and I/O activity but I think it >> makes a terrible Web server or shell machine. Linux is more suitable for >> interactive type work but blows as a network router. > >May I ask by which data you came to your conclusions? >-- Web servers require quick TCP connect times, disk reads and (given use of cgi scripts) fork/exec times. FreeBSD is very good at each. The key is (as in any system with SVR4/SunOS style shared libs) to build frequently executed programs static. My measurements have shown that FreeBSD is on par with Linux for fork/exec when both systems use shared libs (FreeBSD-SunOS, Linux-SVR3), and is faster than Linux for fork/exec when both systems do not. Note that I would expect that Linux-SunOS (ELF) would be even slower. There are some very impressively (big, not just 10K-100K hits/day) large Web sites that use FreeBSD -- very effectively. Also, except under certain circumstances, FreeBSD has very smooth performance (and has had for quite a long time) when using X-windows in an interactive environment. Some changes were made in '94 (a long time ago) to fix the BSD scheduling algorithm especially in the area of interactive response. FWIW, John dyson@freebsd.org