Talk:RAID

Caps
We should use title case for Redundant Array of Inexpensive Disks only if RAID is considered to be a proper noun. If not, redundant array of inexpensive disks is more compatible with the MOS (see WP:EXPABBR and WP:PROPERNAME) despite how sources may style it. ~Kvng (talk) 01:48, 1 May 2020 (UTC)
 * IMO initial capitalization is appropriate under MOS:ACROFIRSTUSE which states, "Unless specified in the "Exceptions" section below, an acronym should be written out in full the first time it is used on a page, followed by the abbreviation in parentheses, e.g. Millennium Development Goals (MDGs) if it is used later in the article." Note the initial capitalization of Millennium Development Goals. RAID is the title of this article and the first time the acronym is written out it follows this policy. I would also note that the spelled out acronym frequently appears with initial caps in relevant literature such at the defining Patterson paper, the RAIDbooks, etc.  Tom94022 (talk) 06:58, 1 May 2020 (UTC)
 * That's true but MOS:ACROFIRSTUSE isn't talking about caps, the section refers to WP:EXPABBR for that. Millennium Development Goals gets title case because it is a proper noun, not because it is an acronym definition. I have started a discussion at Abbreviations to improve our guidance on this. ~Kvng (talk) 13:37, 4 May 2020 (UTC)
 * Then I suggest Redundant Array of Inexpensive Disks is a proper noun per dictionary definition and as described at WP:PROPERNAME in that it is a specific storage technology as defined in the original paper and continued in general use. FWIW I looked for a discussion at Abbreviations but didn't find one.  Tom94022 (talk) 20:01, 4 May 2020 (UTC)
 * Sorry about the bum link. Wikipedia_talk:Manual_of_Style/Abbreviations is the correct link.
 * I don't have a good track record identifying proper nouns for these topics so I'm not going to argue about that if that's now the justification. It does look like a different call was made at Redundant array of independent memory though. ~Kvng (talk) 21:46, 4 May 2020 (UTC)
 * Interesting observation about RAID and RAIM. I suggest that RAID's general acceptance has made its unabbreviated form a proper noun where as RAIM has not yet reached such a level of specificity and industry acceptance to be an unabbreviated proper noun.  I note that RS's use both versions of RAIM but it's main promulgator IBM used the lower case version.  Tom94022 (talk) 00:41, 5 May 2020 (UTC)
 * Style guides used by sources may dictate capitalization in acronym definitions. (Wikipedia's style guide sees things differently in this area.) because of this, seeing title case in a source isn't necessarily evidence that the term is considered a proper noun. ~Kvng (talk) 13:34, 5 May 2020 (UTC)

Unreliable sources
The source 87:

Leventhal, Adam (2009-12-01). "Triple-Parity RAID and Beyond. ACM Queue, Association of Computing Machinery". Retrieved 2012-11-30.

uses uncommon bit error rates, with EVEN FICTIONAL VALUES to proof something that not exists. Raid 6 2019 same reliability as 5 in 2010 ?????

Highly questionable way to calculate. MTBF is used in REALITY, increasing the reliability of bit-error-rates drastically.

92.116.65.204 (talk) 19:54, 24 October 2020 (UTC)

Seems like a student published his homework. 92.116.65.204 (talk) 20:12, 24 October 2020 (UTC)

Unreliable, obsolete source 11
Chen, Peter; Lee, Edward; Gibson, Garth; Katz, Randy; Patterson, David (1994). "RAID: High-Performance, Reliable Secondary Storage". ACM Computing Surveys. 26 (2): 145–185. CiteSeerX 10.1.1.41.3889. doi:10.1145/176979.176981.

26 Years old and same error as above: Bit-error instead of MTBF.

 1014: This would mean a single 12TB SATA drive is only readable/writable once, a 20TB cannot be read/written a SINGLE TIME!!! 92.116.65.204 (talk) 21:32, 24 October 2020 (UTC)

Note on Bit Error Rate
Bit error rates are averages and can be misleading taken by themselves, because errors may not be distributed randomly among the individual bits. The errors may be clustered. This means that the lack of errors could also be clustered. Instead of 1014 good bits followed by 1 bad bit, you might have 1014 good blocks followed by one bad block. The average bit error rate in both these cases is exactly the same, but the latter would be much easier for most block-based RAID implementations to deal with. &mdash; Unfortunately I do not have a good source to cite for this. — Preceding unsigned comment added by 2601:344:8080:A50:ED61:6F21:54DD:728 (talk) 04:40, 28 February 2021 (UTC)

RAID-5 write hole.
I check the 1994 paper mentioned in footnote 11. Is does not contain the word "hole". Your attribution appears to be incorrect. 149.32.192.47 (talk) 17:38, 20 February 2023 (UTC)


 * Thanks for the heads up – I've marked the source as failing verification. The term itself is in use though, we just need a better source. --Zac67 (talk) 18:38, 20 February 2023 (UTC)


 * Fixed? Tom94022 (talk) 21:05, 20 February 2023 (UTC)

Introduced in Windows in 2000 incorrect?
I've been playing around with Windows NT 3.51 in an emulator and the diskmgmt.msc within NT 3.51 supports RAID 0 and 1, though simply called "Striped" and "Mirrored" respectively. The article seems to imply this sort of thing was introduced with dynamic disks in 2000. Should this be corrected? 172.4.121.7 (talk) 13:48, 18 February 2024 (UTC)