From owner-freebsd-hackers Tue Jan 30 07:07:09 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id HAA26986 for hackers-outgoing; Tue, 30 Jan 1996 07:07:09 -0800 (PST) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id HAA26973 for ; Tue, 30 Jan 1996 07:06:58 -0800 (PST) Received: (from root@localhost) by dyson.iquest.net (8.6.11/8.6.9) id KAA03213; Tue, 30 Jan 1996 10:02:46 GMT From: "John S. Dyson" Message-Id: <199601301002.KAA03213@dyson.iquest.net> Subject: Re: intested! To: rajp@nando.net (rajp) Date: Tue, 30 Jan 1996 10:02:46 +0000 () Cc: hackers@freebsd.org In-Reply-To: <199601301312.IAA27751@bessel.nando.net> from "rajp" at Jan 30, 96 08:12:40 am X-Mailer: ELM [version 2.4 PL24 ME8] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-hackers@freebsd.org Precedence: bulk > > Hey, what does that line about linux mean?! > > How about telling me what you guys have in mind on the internals side? > Like any device drivers,etc? > > -Raj > (This is from a -core team member, but only HIS opinion, not that of -core... Some other -core members *might* agree.) FreeBSD is a VERY high performance OS/Kernel. At times, the are specific areas where Linux is faster. However, those times are mostly under light load conditions. Under heavy load, the algorithms backing FreeBSD withstand more use. It isn't just an O(n) type thing. A good example is the paging stuff. Linux doesn't use a statistics based scheme that is very complex, FreeBSD does. Try running random paging benchmarks -- you will then see it clearly, even the recent Linux stuff thrashes very easily. (I have some that I use as regression tests that right now, but I want to keep private.) FreeBSD is faster on the same piece of hardware for file copies and other things (like networking) also. The thing that FreeBSD doesn't have that Linux has is the "cult of personality". The FreeBSD team is made up of older computer professionals (many are more than even the ripe old age of 25.) A couple of us have seen the mistakes in other OSes and are not going to re-create them (at least for long.) FreeBSD does have one major "problem" that I worry about alot, and that is the disk metadata thing. We have been improving the situation, and hopefully soon will make a major performance impact. If the tests at Usenix would have been done with FreeBSD-current, the meta-data operations would have been much faster if the filesystems were mounted -async. FreeBSD initially chooses a more robust disk meta-data update policy, and that had been the recent history of BSD Eventually, our -async option will be better supported, with perhaps a bit of risk. But it will be no more (and perhaps less) risk than the Linux approach. Regarding user-land, please refer to the ports collection, or almost anything that was (user-land) written for any BSD or some SVR4 variants. They usually port cleanly. It is less of an issue with BSD, because it is a bit more standard (historical) than Linux (which is a bit of a mixture of philosophies.) You will definitely find fewer games and toys for BSD though, but our Linux emulation helps that :-). You'll probably soon find FreeBSD running Linux as well as or better than Linux :-). We can already do a pretty good job. As to the future, more performance, more of the 4.4Lite/2 exotic filesystems. More compatibility with other OSes. IPv6, better working NFS. Things will be getting better and better!!! John Dyson