From owner-freebsd-chat Thu May 20 16:59:59 1999 Delivered-To: freebsd-chat@freebsd.org Received: from ns1.yes.no (ns1.yes.no [195.204.136.10]) by hub.freebsd.org (Postfix) with ESMTP id 0C50215009 for ; Thu, 20 May 1999 16:59:53 -0700 (PDT) (envelope-from eivind@bitbox.follo.net) Received: from bitbox.follo.net (bitbox.follo.net [195.204.143.218]) by ns1.yes.no (8.9.1a/8.9.1) with ESMTP id BAA26313; Fri, 21 May 1999 01:59:49 +0200 (CEST) Received: (from eivind@localhost) by bitbox.follo.net (8.8.8/8.8.6) id BAA78487; Fri, 21 May 1999 01:59:48 +0200 (MET DST) Date: Fri, 21 May 1999 01:59:47 +0200 From: Eivind Eklund To: David Kelly Cc: The Hermit Hacker , freebsd-chat@FreeBSD.ORG Subject: Re: SGI, XFS and OSS? Message-ID: <19990521015947.R76043@bitbox.follo.net> References: <199905202101.QAA79579@nospam.hiwaay.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.95.1i In-Reply-To: <199905202101.QAA79579@nospam.hiwaay.net>; from David Kelly on Thu, May 20, 1999 at 04:01:22PM -0500 Sender: owner-freebsd-chat@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Thu, May 20, 1999 at 04:01:22PM -0500, David Kelly wrote: [On FreeBSD w/soft updates being slower than XFS] > Yup. But all measurements were seat-of-the-pants. Didn't much bother to > time things as at the time all that could do was to make me unhappy. An > SGI system could fly right thru tar'ing FreeBSD's ports tree, either on > read from tape or write to tape, where FreeBSD 3.1 with softupdates and > 2.2.8 (without) can't keep the DDS-2 tape drive streaming (400k/sec). > 10k blocksize in both cases. > > At my now former employer, I kept /home/ncvs and /usr/ports hosted on an > SGI O2, 180MHz, 64MB RAM, and let the FreeBSD systems access via 10baseT > ethernet. Mostly because the SGI was where disk space was available. > Partly because it seemed faster. > > Another good test of speed was "rm -rf /usr/ports". The O2 could do it > so fast it was frightening. Heh - both of those are due to problems with the directory layout logic, I think, not due to problems with FFS in itself. The logic for where to put the inode for new directories is very bad for the ports collection; it should be tuned. However, I do not see that as a reason for replacing the entire FS code :-) It is possible XFS will be faster no matter what, but I don't think it will be that much faster. Eivind. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message