From owner-freebsd-fs Fri Jun 14 13:56:18 1996 Return-Path: owner-fs Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id NAA16713 for fs-outgoing; Fri, 14 Jun 1996 13:56:18 -0700 (PDT) Received: from who.cdrom.com (who.cdrom.com [204.216.27.3]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id NAA16688 for ; Fri, 14 Jun 1996 13:56:13 -0700 (PDT) Received: from rocky.sri.MT.net (rocky.sri.MT.net [204.182.243.10]) by who.cdrom.com (8.6.12/8.6.11) with ESMTP id NAA10527 for ; Fri, 14 Jun 1996 13:20:43 -0700 Received: (from nate@localhost) by rocky.sri.MT.net (8.6.12/8.6.12) id OAA22670; Fri, 14 Jun 1996 14:20:35 -0600 Date: Fri, 14 Jun 1996 14:20:35 -0600 From: Nate Williams Message-Id: <199606142020.OAA22670@rocky.sri.MT.net> To: Joe McGuckin Cc: nate@sri.MT.net, freebsd-fs@FreeBSD.ORG Subject: Re: Extent like FFS In-Reply-To: <199606142016.NAA20592@ns.via.net> References: <199606142016.NAA20592@ns.via.net> Sender: owner-fs@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Why not incorporate some ideas from Sprite? One feature that seemed to > speed things up in general was the fact that no metadata gets written out > for a newly created file until it has aged 30 seconds. I think this is a lose as far as POSIX goes, but the FS gurus can thrash this out. > It seems to me this could be a win in a couple of ways: > > - Small, tmp files exist only in the buffer cache & never get > written to disk if they don't live more that 30 seconds. If you are worried about files in /tmp, use MFS. :) > - There out to be a way to queue all the newly created files together > then write them out together at once after they've lived 'X' amount > of time. And when your system power goes away in the middle of writing out these files and they happen to be pretty important what happens? > Where can I find an indepth explanation of how the FFS file system code works? There are papers in the system, plus the 4.4BSD book apparently goes into detail on it. Nate