*BSD News Article 73813


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!nntp.coast.net!lll-winken.llnl.gov!uwm.edu!news.sol.net!brasil.moneng.mei.com!not-for-mail
From: jgreco@brasil.moneng.mei.com (Joe Greco)
Newsgroups: news.software.nntp,comp.unix.bsd.freebsd.misc
Subject: Re: FreeBSD 2.1.5, INN 1.4unoff4, and mmap()
Date: 15 Jul 1996 10:17:10 -0500
Organization: Marquette Electronics, Inc. - Milwaukee, WI.
Lines: 32
Distribution: inet
Message-ID: <4sdndm$27k@brasil.moneng.mei.com>
References: <mandrews.837437077@bob.wittenberg.edu> <4sdk50$9cp@news.demos.su>
NNTP-Posting-Host: brasil.moneng.mei.com
Xref: euryale.cc.adfa.oz.au news.software.nntp:24415 comp.unix.bsd.freebsd.misc:23623

In news.software.nntp article <4sdk50$9cp@news.demos.su>, andy@sun-fox.demos.su (Andrew A. Vasilyev) wrote:
:Mike Andrews (mandrews@bob.Wittenberg.EDU) wrote:
:> Now that FreeBSD 2.1.5-RELEASE is out, is it safe to compile in the use
:> of mmap() in INN 1.4unoff4?  Currently I have ACT_STYLE set to READ, and
:
:  In 2.2-Current (a week before 2.1.5-R) mmap() was broken - nulls
:  were added to 8K boundary at the end of active. Probably it is unsafe
:  to use it in 2.1.5R.
:
:> working, but occasionally I get dangling symlinks scattered across the disk.
:> Maybe I just goofed somewhere, but ccd sounds like it'd be a performance win
:> as well.
:
:  We've tested ccd on 2.2-C, on the 2 2G Seagates the best performance
:  was as of one device (with interleave 64 and 128): read/write were
:  5.0-5.1/2.6-2.7 Mbps, direct i/o (dd if=/dev/rccd0 of=/dev/null) was
:  8.7 Mbps, the same of one disk is above 5 Mbps, so one can see that
:  the performance of raw metadevice is much more than of metadevice FS.

Why would you want to do this on a news server, anyways?

Use a _large_ stripe size to increase available concurrency of accesses.
You don't care about how FAST you can retrieve your 1K or 2K Usenet articles
off of the disk.  You will not derive a benefit from striping from
bandwidth.  You care about how MANY you can get off your disks in as short a
period as possible.  If you have four drives, you would ideally like to be
able to saturate the device and have each drive doing a complete article
fetch without involving other drives.

I use a stripe size of 65536.  And it really works well.

... JG