Date: Sun, 19 May 2013 16:05:36 -0400 From: Paul Kraus <paul@kraus-haus.org> To: kpneal@pobox.com Cc: Ivailo Tanusheff <Ivailo.Tanusheff@skrill.com>, Liste FreeBSD <freebsd-questions@freebsd.org> Subject: Re: ZFS install on a partition Message-ID: <E29A91ED-3B95-402B-B03D-E6C7D319B92B@kraus-haus.org> In-Reply-To: <20130519021642.GC3841@neutralgood.org> References: <F744BBF1-D98C-47BF-9546-14D1A9CB3733@todoo.biz> <372082cab2064846809615a8073e022c@DB3PR07MB059.eurprd07.prod.outlook.com> <A9599DD7-1A32-4607-BC83-2E6E4D03C560@kraus-haus.org> <b179c20ebde742358e2cc52a1f04133e@DB3PR07MB059.eurprd07.prod.outlook.com> <20130519021642.GC3841@neutralgood.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On May 18, 2013, at 10:16 PM, kpneal@pobox.com wrote: > On Sat, May 18, 2013 at 01:29:58PM +0000, Ivailo Tanusheff wrote: >> Not sure about your calculations, hope you trust them, but in my = previous company we have a 3-4 months period when a disk fails almost = every day on 2 year old servers, so trust me - I do NOT trust those = calculations, as I've seen the opposite. Maybe it was a failed batch of = disk, shipped in the country, but no one is insured against this. Yes, = you can use several hot spares on the software raid, but: >=20 > What calculations are you talking about? He posted the uncorrectable = read > error probabilities manufacturers put into drive datasheets. The = probability > of a URE is distinct from and very different from the probability of = the > entire drive failing. I think he is referring to the calculation I did based on = uncorrectable error rate and whether you will run into that type of = error over the life of the drive. 1 TB =3D=3D 8,796,093,022,208 bits 10^15 (in bits) / 1 TB ~=3D 113.687 So if over the life of the drive you READ a TOTAL of 113.687 TB, = then you will, statistically speaking, run into one uncorrectable read = error and potentially return bad data to the application or OS. This = does NOT scale with size of drive, it is the same for all drives with an = uncorrectable error rate of 10^-15 bits. So if you read the entirety of = a 1 TB drive 114 times or a 4 TB 29 times you get the same result. But this is a statistical probability, and some drives will have = more (much more) uncorrectable errors and others will have less (much = less), although I don't know if the distribution falls on a typical = gaussian (bell) curve. -- Paul Kraus Deputy Technical Director, LoneStarCon 3 Sound Coordinator, Schenectady Light Opera Company
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E29A91ED-3B95-402B-B03D-E6C7D319B92B>