From owner-freebsd-database Sun Apr 14 12:18:16 2002 Delivered-To: freebsd-database@freebsd.org Received: from elvis.mu.org (elvis.mu.org [192.203.228.196]) by hub.freebsd.org (Postfix) with ESMTP id 88F1037B405 for ; Sun, 14 Apr 2002 12:18:14 -0700 (PDT) Received: by elvis.mu.org (Postfix, from userid 1061) id 5CB79AE1D1; Sun, 14 Apr 2002 12:18:14 -0700 (PDT) Date: Sun, 14 Apr 2002 12:18:14 -0700 From: David Drum To: FreeBSD DB List Subject: Re: Raid configuration Message-ID: <20020414191814.GC40226@elvis.mu.org> Mail-Followup-To: David Drum , FreeBSD DB List References: <20020411235022.GA21045@elvis.mu.org> <20020412120947.Q48494-100000@mail.wolves.k12.mo.us> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20020412120947.Q48494-100000@mail.wolves.k12.mo.us> User-Agent: Mutt/1.3.27i Sender: owner-freebsd-database@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Quoth Chris Dillon: > And when you only have a four-drive configuration, it makes no > difference which one you use since the chances of a total failure > is exactly the same either way. Any more drives than that and you > definately want RAID10. :-) The chances of total failure may be the same, but the effort required to rebuild the RAID is not. If you have 4 9GB disks in a RAID 0+1 and one goes bad, you have to mirror 18GB once the drive is replaced. If you have a RAID 1+0, you only have one drive to mirror, and not a stripe. So RAID 1+0 has clear advantages even in the minimal implementation. Regards, David Drum david@mu.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-database" in the body of the message From owner-freebsd-database Sun Apr 14 14:19:24 2002 Delivered-To: freebsd-database@freebsd.org Received: from elvis.mu.org (elvis.mu.org [192.203.228.196]) by hub.freebsd.org (Postfix) with ESMTP id B128837B404 for ; Sun, 14 Apr 2002 14:19:20 -0700 (PDT) Received: by elvis.mu.org (Postfix, from userid 1061) id 90F1DAE162; Sun, 14 Apr 2002 14:19:20 -0700 (PDT) Date: Sun, 14 Apr 2002 14:19:20 -0700 From: David Drum To: FreeBSD DB List Subject: Re: Raid configuration Message-ID: <20020414211920.GA42961@elvis.mu.org> Mail-Followup-To: David Drum , FreeBSD DB List References: <20020329113017.L51218-100000@zoraida.natserv.net> <20020411180923.O56810@flake.decibel.org> <20020411235022.GA21045@elvis.mu.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20020411235022.GA21045@elvis.mu.org> User-Agent: Mutt/1.3.27i Sender: owner-freebsd-database@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Whoops, The daemon of cut and paste "cutnpasted(8)" struck in my original post. The first comment below should be: RAID 1+0 (a stripe (RAID 0) of mirrored (RAID 1) drives) My apologies for the confusion. Quoth David Drum: > stripe > +-------------+ > | mirror1 | > |+-----------+| > || 1a 1b || RAID 0+1 (a mirror (RAID 0) of striped (RAID 1) drives) > |+-----------+| > | mirror2 | If any one disk in any of the three mirrors goes bad, > |+-----------+| the stripe is still preserved. > || 2a 2b || > |+-----------+| > | mirror3 | > |+-----------+| > || 3a 3b || > |+-----------+| > +-------------+ Regards, David Drum david@mu.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-database" in the body of the message From owner-freebsd-database Sun Apr 14 19:29:30 2002 Delivered-To: freebsd-database@freebsd.org Received: from mail.wolves.k12.mo.us (mail.wolves.k12.mo.us [207.160.214.1]) by hub.freebsd.org (Postfix) with ESMTP id EC10E37B404 for ; Sun, 14 Apr 2002 19:29:26 -0700 (PDT) Received: from mail.wolves.k12.mo.us (cdillon@mail.wolves.k12.mo.us [207.160.214.1]) by mail.wolves.k12.mo.us (8.9.3/8.9.3) with ESMTP id VAA83947; Sun, 14 Apr 2002 21:29:15 -0500 (CDT) (envelope-from cdillon@wolves.k12.mo.us) Date: Sun, 14 Apr 2002 21:29:15 -0500 (CDT) From: Chris Dillon To: David Drum Cc: FreeBSD DB List Subject: Re: Raid configuration In-Reply-To: <20020414191814.GC40226@elvis.mu.org> Message-ID: <20020414212607.X83685-100000@mail.wolves.k12.mo.us> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-database@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG On Sun, 14 Apr 2002, David Drum wrote: > > And when you only have a four-drive configuration, it makes no > > difference which one you use since the chances of a total failure > > is exactly the same either way. Any more drives than that and you > > definately want RAID10. :-) > > The chances of total failure may be the same, but the effort > required to rebuild the RAID is not. If you have 4 9GB disks in a > RAID 0+1 and one goes bad, you have to mirror 18GB once the drive is > replaced. If you have a RAID 1+0, you only have one drive to > mirror, and not a stripe. Ah, yes, I hadn't thought of what it would take to rebuild one. In that case, RAID 0+1 looks like the loser in all situations. -- Chris Dillon - cdillon@wolves.k12.mo.us - cdillon@inter-linc.net FreeBSD: The fastest and most stable server OS on the planet - Available for IA32 (Intel x86) and Alpha architectures - IA64, PowerPC, UltraSPARC, and ARM architectures under development - http://www.freebsd.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-database" in the body of the message From owner-freebsd-database Tue Apr 16 9:18:43 2002 Delivered-To: freebsd-database@freebsd.org Received: from kali.avantgo.com (shadow.avantgo.com [64.157.226.66]) by hub.freebsd.org (Postfix) with ESMTP id 06C9037B419 for ; Tue, 16 Apr 2002 09:18:38 -0700 (PDT) Received: from river.avantgo.com ([10.11.30.114]) by kali.avantgo.com with Microsoft SMTPSVC(5.0.2195.3779); Tue, 16 Apr 2002 09:18:37 -0700 Date: Tue, 16 Apr 2002 09:19:01 -0700 (PDT) From: Scott Hess To: Chris Dillon Cc: David Drum , FreeBSD DB List Subject: Re: Raid configuration In-Reply-To: <20020414212607.X83685-100000@mail.wolves.k12.mo.us> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-OriginalArrivalTime: 16 Apr 2002 16:18:37.0844 (UTC) FILETIME=[5D543540:01C1E562] Sender: owner-freebsd-database@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Additionally, consider two setups: RAID10 (stripe of mirror) +---------+ |+-------+| || A = B || |+-------+| |+-------+| || C = D || |+-------+| +---------+ RAID01 (mirror of stripes) +-----------+ |+---+ +---+| || A | | B || || |=| || || C | | D || |+---+ +---+| +-----------+ Both have the same uptime for single-disk failures. For two-disk failures, RAID10 stays up for 2/3 of the cases, while RAID01 only stays up in 1/3 of the cases. Later, scott On Sun, 14 Apr 2002, Chris Dillon wrote: > On Sun, 14 Apr 2002, David Drum wrote: > > > > And when you only have a four-drive configuration, it makes no > > > difference which one you use since the chances of a total failure > > > is exactly the same either way. Any more drives than that and you > > > definately want RAID10. :-) > > > > The chances of total failure may be the same, but the effort > > required to rebuild the RAID is not. If you have 4 9GB disks in a > > RAID 0+1 and one goes bad, you have to mirror 18GB once the drive is > > replaced. If you have a RAID 1+0, you only have one drive to > > mirror, and not a stripe. > > Ah, yes, I hadn't thought of what it would take to rebuild one. In > that case, RAID 0+1 looks like the loser in all situations. > > -- > Chris Dillon - cdillon@wolves.k12.mo.us - cdillon@inter-linc.net > FreeBSD: The fastest and most stable server OS on the planet > - Available for IA32 (Intel x86) and Alpha architectures > - IA64, PowerPC, UltraSPARC, and ARM architectures under development > - http://www.freebsd.org > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-database" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-database" in the body of the message From owner-freebsd-database Tue Apr 16 9:40: 0 2002 Delivered-To: freebsd-database@freebsd.org Received: from kali.avantgo.com (shadow.avantgo.com [64.157.226.66]) by hub.freebsd.org (Postfix) with ESMTP id 787FC37B405 for ; Tue, 16 Apr 2002 09:39:56 -0700 (PDT) Received: from river.avantgo.com ([10.11.30.114]) by kali.avantgo.com with Microsoft SMTPSVC(5.0.2195.3779); Tue, 16 Apr 2002 09:39:55 -0700 Date: Tue, 16 Apr 2002 09:40:19 -0700 (PDT) From: Scott Hess To: FreeBSD DB List Subject: Re: Raid configuration In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-OriginalArrivalTime: 16 Apr 2002 16:39:55.0377 (UTC) FILETIME=[56CC4610:01C1E565] Sender: owner-freebsd-database@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG So, why'd I draw the diagram if I wasn't going to refer to the disks by name at any point? If 2 disks fail, there are six possible pairs. RAID10 stays up if (A|B)&(C|D) stay up (4/6). RAID01 can only stay up if (A&C)|(B&D) stay up (2/6). Sigh, scott On Tue, 16 Apr 2002, Scott Hess wrote: > Additionally, consider two setups: > > RAID10 (stripe of mirror) > +---------+ > |+-------+| > || A = B || > |+-------+| > |+-------+| > || C = D || > |+-------+| > +---------+ > > RAID01 (mirror of stripes) > +-----------+ > |+---+ +---+| > || A | | B || > || |=| || > || C | | D || > |+---+ +---+| > +-----------+ > > Both have the same uptime for single-disk failures. For two-disk > failures, RAID10 stays up for 2/3 of the cases, while RAID01 only stays up > in 1/3 of the cases. > > Later, > scott > > > On Sun, 14 Apr 2002, Chris Dillon wrote: > > On Sun, 14 Apr 2002, David Drum wrote: > > > > > > And when you only have a four-drive configuration, it makes no > > > > difference which one you use since the chances of a total failure > > > > is exactly the same either way. Any more drives than that and you > > > > definately want RAID10. :-) > > > > > > The chances of total failure may be the same, but the effort > > > required to rebuild the RAID is not. If you have 4 9GB disks in a > > > RAID 0+1 and one goes bad, you have to mirror 18GB once the drive is > > > replaced. If you have a RAID 1+0, you only have one drive to > > > mirror, and not a stripe. > > > > Ah, yes, I hadn't thought of what it would take to rebuild one. In > > that case, RAID 0+1 looks like the loser in all situations. > > > > -- > > Chris Dillon - cdillon@wolves.k12.mo.us - cdillon@inter-linc.net > > FreeBSD: The fastest and most stable server OS on the planet > > - Available for IA32 (Intel x86) and Alpha architectures > > - IA64, PowerPC, UltraSPARC, and ARM architectures under development > > - http://www.freebsd.org > > > > > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > > with "unsubscribe freebsd-database" in the body of the message > > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-database" in the body of the message From owner-freebsd-database Tue Apr 16 10:27:44 2002 Delivered-To: freebsd-database@freebsd.org Received: from ws2.hk5.outblaze.com (202-77-181-84.outblaze.com [202.77.181.84]) by hub.freebsd.org (Postfix) with SMTP id 43BD937B404 for ; Tue, 16 Apr 2002 10:27:34 -0700 (PDT) Received: (qmail 31991 invoked by uid 1001); 16 Apr 2002 17:27:27 -0000 Message-ID: <20020416172727.31988.qmail@graffiti.net> Content-Type: text/plain; charset="iso-8859-1" Content-Disposition: inline Content-Transfer-Encoding: 7bit Mime-Version: 1.0 X-Mailer: MIME-tools 4.104 (Entity 4.117) Received: from [66.51.217.108] by ws2.hk5.outblaze.com with http for tedstriker@graffiti.net; Wed, 17 Apr 2002 01:27:27 +0800 From: "Ted Striker" To: , Date: Wed, 17 Apr 2002 01:27:27 +0800 Subject: Re: Raid configuration X-Originating-Ip: 66.51.217.108 X-Originating-Server: ws2.hk5.outblaze.com Sender: owner-freebsd-database@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG I know this was a question regarding Vinum, but, our RAID card only let us set up a RAID10, there was no option to do RAID01. I think that says something about which one is more reliable. ----- Original Message ----- From: Scott Hess Date: Tue, 16 Apr 2002 09:40:19 -0700 (PDT) To: FreeBSD DB List Subject: Re: Raid configuration > So, why'd I draw the diagram if I wasn't going to refer to the disks by > name at any point? If 2 disks fail, there are six possible pairs. > RAID10 stays up if (A|B)&(C|D) stay up (4/6). RAID01 can only stay up if > (A&C)|(B&D) stay up (2/6). > > Sigh, > scott > > On Tue, 16 Apr 2002, Scott Hess wrote: > > Additionally, consider two setups: > > > > RAID10 (stripe of mirror) > > +---------+ > > |+-------+| > > || A = B || > > |+-------+| > > |+-------+| > > || C = D || > > |+-------+| > > +---------+ > > > > RAID01 (mirror of stripes) > > +-----------+ > > |+---+ +---+| > > || A | | B || > > || |=| || > > || C | | D || > > |+---+ +---+| > > +-----------+ > > > > Both have the same uptime for single-disk failures. For two-disk > > failures, RAID10 stays up for 2/3 of the cases, while RAID01 only stays up > > in 1/3 of the cases. > > > > Later, > > scott > > > > > > On Sun, 14 Apr 2002, Chris Dillon wrote: > > > On Sun, 14 Apr 2002, David Drum wrote: > > > > > > > > And when you only have a four-drive configuration, it makes no > > > > > difference which one you use since the chances of a total failure > > > > > is exactly the same either way. Any more drives than that and you > > > > > definately want RAID10. :-) > > > > > > > > The chances of total failure may be the same, but the effort > > > > required to rebuild the RAID is not. If you have 4 9GB disks in a > > > > RAID 0+1 and one goes bad, you have to mirror 18GB once the drive is > > > > replaced. If you have a RAID 1+0, you only have one drive to > > > > mirror, and not a stripe. > > > > > > Ah, yes, I hadn't thought of what it would take to rebuild one. In > > > that case, RAID 0+1 looks like the loser in all situations. > > > > > > -- > > > Chris Dillon - cdillon@wolves.k12.mo.us - cdillon@inter-linc.net > > > FreeBSD: The fastest and most stable server OS on the planet > > > - Available for IA32 (Intel x86) and Alpha architectures > > > - IA64, PowerPC, UltraSPARC, and ARM architectures under development > > > - http://www.freebsd.org > > > > > > > > > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > > > with "unsubscribe freebsd-database" in the body of the message > > > > > > > > > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-database" in the body of the message > -- _______________________________________________ Get your free email from http://www.graffiti.net Powered by Outblaze To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-database" in the body of the message From owner-freebsd-database Tue Apr 16 13:35:54 2002 Delivered-To: freebsd-database@freebsd.org Received: from mail.wolves.k12.mo.us (mail.wolves.k12.mo.us [207.160.214.1]) by hub.freebsd.org (Postfix) with ESMTP id C57B337B405 for ; Tue, 16 Apr 2002 13:35:50 -0700 (PDT) Received: from mail.wolves.k12.mo.us (cdillon@mail.wolves.k12.mo.us [207.160.214.1]) by mail.wolves.k12.mo.us (8.9.3/8.9.3) with ESMTP id PAA21584; Tue, 16 Apr 2002 15:35:46 -0500 (CDT) (envelope-from cdillon@wolves.k12.mo.us) Date: Tue, 16 Apr 2002 15:35:46 -0500 (CDT) From: Chris Dillon To: Scott Hess Cc: David Drum , FreeBSD DB List Subject: Re: Raid configuration In-Reply-To: Message-ID: <20020416152522.E12575-100000@mail.wolves.k12.mo.us> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-database@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG On Tue, 16 Apr 2002, Scott Hess wrote: > Both have the same uptime for single-disk failures. For two-disk > failures, RAID10 stays up for 2/3 of the cases, while RAID01 only > stays up in 1/3 of the cases. Hmm, yes, that is the case as well... Based on your example diagram, there are six possible two-disk failure cases: Fails RAID10 RAID0+1 ===== ====== ======= A&B BAD BAD A&C OK OK A&D OK BAD B&C OK BAD B&D OK OK C&D BAD BAD As you said, 2/3 chance of surviving with RAID10, and only 1/3 for RAID0+1. -- Chris Dillon - cdillon@wolves.k12.mo.us - cdillon@inter-linc.net FreeBSD: The fastest and most stable server OS on the planet - Available for IA32 (Intel x86) and Alpha architectures - IA64, PowerPC, UltraSPARC, and ARM architectures under development - http://www.freebsd.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-database" in the body of the message