From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-oi0-f48.google.com ([209.85.218.48]:35423 "EHLO mail-oi0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750721AbcFDTNm (ORCPT ); Sat, 4 Jun 2016 15:13:42 -0400 Received: by mail-oi0-f48.google.com with SMTP id w184so173541080oiw.2 for ; Sat, 04 Jun 2016 12:13:41 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <1465061839.6689.9.camel@scientia.net> References: <1464819934.6742.71.camel@scientia.net> <1464975482.6679.11.camel@scientia.net> <6f18c0d1-8ac5-c325-0ba8-ffb949c54554@gmail.com> <1465005092.6648.39.camel@scientia.net> <57528238.6080809@gmail.com> <1465061839.6689.9.camel@scientia.net> Date: Sat, 4 Jun 2016 13:13:40 -0600 Message-ID: Subject: Re: btrfs From: Chris Murphy To: Christoph Anton Mitterer Cc: Chris Murphy , Austin S Hemmelgarn , Btrfs BTRFS Content-Type: text/plain; charset=UTF-8 Sender: linux-btrfs-owner@vger.kernel.org List-ID: On Sat, Jun 4, 2016 at 11:37 AM, Christoph Anton Mitterer wrote: > On Sat, 2016-06-04 at 11:00 -0600, Chris Murphy wrote: >> SNIA's DDF 2.0 spec Rev 19 >> page 18/19 shows 'RAID-1 Simple Mirroring" vs "RAID-1 Multi- >> Mirroring" > > And DDF came how many years after the original RAID paper and everyone > understood RAID1 as it was defined there? 1987 vs. ~2003 or so? > > Also SINA's "standard definition" seems pretty strange, doesn't it? > They have two RAID1, as you say: > - "simple mirroring" with n=2 > - "multi mirrioring" with n=3 > > I wouldn't see why the n=2 case is "simpler" than the n=3 case, neither > why the n=3 case is multi and the n=2 is not (it's also already > multiple disks). > Also why did they allow n=3 but not n>=3? If n=4 wouldn't make sense, > why would n=3, compared to n=2? > > Anyway,... > - the original paper defines it as n mirrored disks > - Wikipedia handles it like that > - the already existing major RAID implementation (MD) in the Linux > kernel handles it like that > - LVM's native mirroring, allows to set the number of mirrors, i.e. it > allows for everything >=2 which is IMHO closer to the common meaning > of RAID1 than to btrfs' two-duplicates > > So even if there would be some reasonable competing definition (and I > don't think the rather proprietary DDF is very reasonable here), why > using one that is incomptabible with everything we have in Linux? mdadm supports DDF. -- Chris Murphy