From owner-freebsd-hackers Wed Oct 21 18:12:54 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id SAA07497 for freebsd-hackers-outgoing; Wed, 21 Oct 1998 18:12:54 -0700 (PDT) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from allegro.lemis.com (allegro.lemis.com [192.109.197.134]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id SAA07490 for ; Wed, 21 Oct 1998 18:12:50 -0700 (PDT) (envelope-from grog@freebie.lemis.com) Received: from freebie.lemis.com (freebie.lemis.com [192.109.197.137]) by allegro.lemis.com (8.9.1/8.9.0) with ESMTP id KAA21490; Thu, 22 Oct 1998 10:42:12 +0930 (CST) Received: (from grog@localhost) by freebie.lemis.com (8.9.1/8.9.0) id KAA19281; Thu, 22 Oct 1998 10:42:10 +0930 (CST) Message-ID: <19981022104209.G1219@freebie.lemis.com> Date: Thu, 22 Oct 1998 10:42:09 +0930 From: Greg Lehey To: Peter Jeremy , hackers@FreeBSD.ORG Subject: Re: FS Behaviour with small frags References: <98Oct22.093607est.40343@border.alcanet.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.91.1i In-Reply-To: <98Oct22.093607est.40343@border.alcanet.com.au>; from Peter Jeremy on Thu, Oct 22, 1998 at 09:36:35AM +1000 WWW-Home-Page: http://www.lemis.com/~grog Organization: LEMIS, PO Box 460, Echunga SA 5153, Australia Phone: +61-8-8388-8286 Fax: +61-8-8388-8725 Mobile: +61-41-739-7062 Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Thursday, 22 October 1998 at 9:36:35 +1000, Peter Jeremy wrote: > I'm looking at using a 4096/512 filesystem (instead of the standard > 8192/1024) for news spool. (Based on an analysis of my current news > spool, this will save about 10% of the space). > > Are there any down-sides to using 4096/512 instead of 8192/1024? For > the same disk and cylinder group organisation, will a 4096/512 FS be > slower or use more system resources than an 8192/1024 FS (given lots > of small files)? Yes, both. Of course, the main system resource is inodes, and you obviously want more of them. The speed problem is mainly a matter of greater fragmentation, of which I would expect a lot in a news server. > My major concerns are that either the buffer cache will allocate 8K > buffers and not use half of them (effectively halving the cache > size), No, that won't happen. > or that allocating a mixture of 4K and 8K buffers will lead to > excessive fragmentation (reducing the performance of the remaining > 8K FS). No, the size of the buffer isn't necessarily the block size of the file system. ufs is quite clever in assigning optimum buffer sizes. I've been examining the request size for Vinum, and I find request (=block) sizes between 512 and 61440 bytes. When copying large files, for example, you're liable to see the 60 kB buffers. Greg -- See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message