From owner-freebsd-questions Fri Sep 3 16:55: 0 1999 Delivered-To: freebsd-questions@freebsd.org Received: from camelot.bitart.com (camelot.BITart.com [206.103.221.33]) by hub.freebsd.org (Postfix) with SMTP id 8D3D2150A9 for ; Fri, 3 Sep 1999 16:54:57 -0700 (PDT) (envelope-from gerti@bitart.com) Received: (qmail 9688 invoked by uid 101); 3 Sep 1999 23:55:01 -0000 Message-ID: <19990903235501.9687.qmail@camelot.bitart.com> Content-Type: text/plain MIME-Version: 1.0 (NeXT Mail 4.2mach v148) In-Reply-To: <19990904090824.Y95378@freebie.lemis.com> X-Nextstep-Mailer: Mail 4.2mach (Enhance 2.2p1) Received: by NeXT.Mailer (1.148) From: Gerd Knops Date: Fri, 3 Sep 1999 18:55:00 -0500 To: Greg Lehey Subject: Re: Vinum: Best performance from mirror setup with two disks Cc: questions@FreeBSD.ORG Reply-To: gerti@BITart.com References: <19990903232137.9612.qmail@camelot.bitart.com> <19990904090824.Y95378@freebie.lemis.com> Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Greg Lehey wrote: > On Friday, 3 September 1999 at 18:21:36 -0500, Gerd Knops wrote: > > Hi, > > > > I have read everything I could find about vinum (man pages, recent > > questions Mailing list, Vinum intro) but I am still not quite sure: > > > > What is the best performing setup to get mirroring with two disks? > > > > I understand that I just set up one plex per disk. Write performance goes > > down due to double writes, read performance goes up since vinum will > > alternate reads between the plexis. > > Correct. > > > What I don't quite understand is if striping gives me an added advantage > > in this scenario. > > It depends a lot on your usage. In general, a striped plex will give > more even access to the two drives, so the performance is likely to be > better. > I still don't get that. In the case of a two disk mirror, how can striping help performance if reads are already alternated between the two plexes (which are on different disks)? Gerd To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message