Talk:Hard disk drive/Archive 11

Consider updates to (or splitting off) components section
I was reviewing sources for our new performance characteristics article and I ran across this article and in this particular section. It has a great logical flow for the inner workings of HDDs. I don't mean to say we should copy the article, but it was very helpful to see the list of key components listed with basic definitions. If we reorganized our components section (maybe bring it up one heading level) I think we could then create links to these component terms in Wikipedia to this section until someone decides it makes sense to write a whole article around some of the components. I will consider adding a more detailed component section, unless we think it should be its own article (which I am inclined to recommend anyway like we are doing for the performance characteristics section. &sect; Music Sorter &sect;  (talk) 05:58, 5 July 2011 (UTC)

Merging from Spindle (computer)
I have already incorporated all the material in the separate Spindle (computer) article into this one including the drawing. This article has much more information surrounding the topic as well. &sect; Music Sorter &sect;  (talk) 22:03, 9 July 2011 (UTC)


 * Support as proposer. &sect; Music Sorter &sect;  (talk) 05:36, 12 July 2011 (UTC)


 * Support the article is quite short and it doesn't seem that too much additional information could be added. However, a redirect should be kept in case a reader wants to access directly this concept. Regards, DPdH (talk) 16:44, 14 July 2011 (UTC)


 * Support the article is about HDD's spindles and is misnamed Spindle (computer). It adds nothing over this article; it should be nuked and the link in Spindle redirected to this article.  Tom94022 (talk) 21:30, 14 July 2011 (UTC)

Dead References
Reference 63 is now 404. Please find additional sources to support the notion of a capacity ceiling. — Preceding unsigned comment added by 46.7.165.130 (talk) 22:14, 10 December 2011 (UTC)
 * I used the site search facility and repaired the link. It is common practice to either tag dead links with dead link, or make a good faith attempt to find out if the page was moved and update the link. But thanks for letting us know. What really needs to be done is permanent archiving of pages that are likely to disappear. This procedure is explained at Using WebCite. Elizium23 (talk) 02:45, 11 December 2011 (UTC)

"In machine-readable data storage..."
Referring to this edit and comment, I think "machine-readable data storage" will be more puzzling to the hypothetical farmer from Africa than "hard disk drive" without a qualifier. It also struck me that the lede took a long time to get around to stating what a HDD is for. (My favorite uncivil editor is fond of ridiculing articles that never, or at least take a long time to, point out what the subject of the article is used for.) By adding "for storing and retrieving ..." to the end of the first sentence, we address both issues in a way much more quickly understood, not to mention far less awkward and stilted, than "In machine-readable data storage...".

Yes, there are uses for hard drives other than computers (consider the Ampex HS-100, which used a hard drive to store analog signals, and with the head contacting the surface besides) but again, it's just the lede. It's enough to say "digital information, primarily computer data". Examples of other uses can appear later. We don't need to mention the occasional analog hard drive here either. It is not the purpose of the lede to describe every exception and special case.

Speaking of which, I also removed some unnecessary detail (and a huge run-on sentence) from the lede. Jeh (talk) 00:17, 16 December 2011 (UTC)
 * I agree with Jeh's edits. Tom94022 (talk) 06:43, 16 December 2011 (UTC)

More on the lede
Tom94022, re your recent change to the lede... I appreciate the updated info but I think that's really too much very deep detail for the lede. Jeh (talk) 07:32, 16 December 2011 (UTC)
 * Hi Jeh, Are you talking about this change to the lead subsection of the Technology section? If so, I think I made it less technical by taking out MFM, GCR, etc and distinguishing between user data and the data as it is encoded onto the disk.  I thought about using Morse code as an analogy but thought that would be TMI.  My problems with the original version were the added tape analogy (IMO not a good one), the several obsolete codes that are just a form of RLL, the way recording was described (sequential changes in polarity versus transitions) and the lack of distinction between encoding and decoding, all of which I tried to fix.  Take a hack at it yourself but please don't go backwards :-)  Tom94022 (talk) 19:32, 17 December 2011 (UTC)

Rewrite Integrity Section
The Integrity section of this article has little to do with Integrity, something about Reliability and a lot about Head Crashes and their mitigation and they are all mashed together with little flow. The recent unsourced addition on [Modes of failure] didn't help. It needs a rewrite, rename and or split. I'll make a proposal here and perhaps sandbox the changes but I thought I would ask for comments before I do major surgery Tom94022 (talk) 19:43, 17 December 2011 (UTC) I thought about this over night and propose that the section be retitled Reliability with the following subsections
 * 7. Reliability
 * Metrics
 * Modes of failure
 * Historical examples
 * Recovery from failures

All of the above would be developed from reliable sources such as the Google study, several papers on HDD reliability such those by Elerath, and web material from HDD companies and HDD recovery services. The stuff on head load/unload and landing zones would be moved to section 2. Any comments Tom94022 (talk) 17:41, 18 December 2011 (UTC) I am now working on this in a sandbox. Contributions are welcome Tom94022 (talk) 19:47, 18 December 2011 (UTC)


 * If you look around on the websites of data recovery firms (bit dubious if those could be used as reliable sources), firmware corruption, especially corruption of the P-LIST and G-LIST, caused by firmware bugs seems to be a common cause for failure of modern drives. This isn't covered yet. —Ruud 23:35, 29 December 2011 (UTC)

using 4gb ram to a 500gb hard disk causes hard disk burn
hello i want to know that,using a 4gb ram to a 500gb hard disk causes the hdd to burn or not. my brother who had to be using a 4 gb ram to a 500gb hdd which burned a lot of times,thanks to the warranty,He got a new ones every time,so in addition he used to be using a cable connection to the desktop for tv,that time if the hdd burns he use to think that the voltage present in the cable cause that burning.and now he upgraded his hdd to 1tb now there is no burning of the hdd of the same company please help me with this — Preceding unsigned comment added by 117.200.98.44 (talk) 19:29, 29 December 2011 (UTC)


 * Highly unlikely. Likely causes include a bad power supply, a bad model hard disk (hard disk manufacturers sometimes produce models with above average failure rates), or just plain bad luck. In the future, please consider posting such questions at the Reference desk. —Ruud 20:06, 29 December 2011 (UTC)


 * IMO it is not possible for the DRAM to cause any failure of an HDD, period, much less causing it to "burn". Since the system now works with the same 4 GiB of RAM, the same power supply, the same cables and a new drive from the same manufacturer it sounds like just plain bad luck.  The cable connections are the only thing that really changed as the drives changed and it is hard to see how they could cause a drive to "burn" - for example, I doubt if an intermittent power supply ground pin or wire would cause a drive to "burn". Likewise the drives are pretty robust so an intermittent power supply is liable to do a lot more damage to other electronics long before it "burns" the drive  I vote for bad luck.  Tom94022 (talk) 22:06, 29 December 2011 (UTC)

Hard disk drive "too technical"?
173.180.88.123 (talk) 23:40, 31 December 2011 (UTC)I would like to comment on the heading that states:

"This article may be too technical for most readers to understand. Please help improve this article to make it understandable to non-experts, without removing the technical details.  The talk page may contain suggestions.  (October 2011)"

This is a technical article, and is therefore "technical". As politically incorrect as it may be, there are many topics that when properly presented, will be beyond the scope of understanding for many readers. I found it quite interesting, there are numerous links, and if I don't understand something, I'll just have to look it up.

All disciplines of study have their own language or terminology that have a specific meaning for the discipline, but a very different meaning to those who have not been educated in the discipline. Replacing this specific terminology or language with simpler language so it becomes more "understandable" will only create ambiguity, and confusion for those who do understand the terminology.

There are no short cuts to learning or understanding.

I do not consider myself a "techie" or "geek", but the vast majority of computer users have very little knowledge about their systems. Some of my friends consider me an "expert", but compared to the real experts I have known over the years, I am not much more than a "newbie".

One of the best examples of misunderstood terminology can be drawn from statistics; accuracy and precision have very different meanings, but they are used interchangeably in common usage.

Trying to “dumb down” information or knowledge to the lowest common denominator leads to mediocrity, and is a disservice to those who have taken considerable time, often at considerable expense, to acquire their expertise.

Wikipedia should also be a source of information for those of us who willing to stretch our minds to acquire more knowledge.

Pablum is not the solution.

jloimand@telus.net


 * Agreed, we should take the tag down. If the IP tag bomber neither took the time to explain why he/she thought the article was "too technical" nor made any attempt to clean up examples then we shouldn't bother to give him/her too much credence.  Tom94022 (talk) 23:48, 31 December 2011 (UTC)


 * Agreed and done. The IP's other contributions do not lend much credence to him or her either. Jeh (talk) 02:24, 1 January 2012 (UTC)

Failure modes and drive orientation
User Johnnynolan2 (contribs, talk) added a claim that vertically orienting a hard drive could contribute to failure. I have reverted this for these reasons:


 * All Wikipedia content must be verifiable by being referenced to reliable sources. There was a source, but it's a self-published blog; those don't count.
 * The claim is contradicted by information from hard drive manufacturers. They do recommend that the platters be within a few degrees of horizontal or vertical, but there is no recommendation against vertical.
 * The claim on the referenced blog is unlikely in that it depends on the access arms simply moving in and out (like a piston). Rather, the arm assembly rotates around a pivot, in the manner of the tone arm on a record turntable. And, like a turntable arm, it is counterweighted on the other side of the pivot. Due to the counterweight, a vertical orientation does not cause gravity to pull the arms in or out. (Hey, the manufacturers did anticipate the possibility of vertical orientation!)
 * The exact arm position is determined by the drive reading signal from the tracks and micro-positioning the heads for the best signal. The drives do not simply command the heads to move to a position and hope for the best. They used to do that (a couple of decades ago, using stepper motors) but no more. So even if gravity did exert an influence on the head position, it would be corrected by the drive.

I hope this adequately explains why I reverted these edits. Further discussion here, please, if anyone wants. Jeh (talk) 06:05, 7 January 2012 (UTC)


 * I've seen some pretty convincing diagrams that imply vertical mounting can significantly raise the number of seek errors (which would reduce performance, but not cause failure), but I don't think those were published on a reputable tech site. —Ruud 16:53, 7 January 2012 (UTC)


 * I agree with Jeh's analysis and support the removal. In modern HDDs the rotary actuators are balanced and the closed loop positioning mechanism can easily compensate for any residual force caused by any residual unbalance.  I doubt if vertical mounting has any effect upon seek errors in modern drives.  Tom94022 (talk) 17:54, 7 January 2012 (UTC)


 * I found an interesting forum thread . According to manufacturers any the 6 orientations is fine although they usually make no comment on tilted ones or say not to mount them tilted (Hitachi). The thread itself is of course not a reliable source, but some of the sources they bases their conclusions on are. We should probably add something on this to the article. —Ruud 18:59, 7 January 2012 (UTC)


 * I suppose a single sentence someplace in the article that modern disk drives can be mounted in any orientation citing the references in the thread would be OK although I am not sure it adds much to this already overlong article.  Regarding tilt - ever seen a prohibition on operating your laptop on your lap?  FWIW, I'm typing this on an IBM T23 in an IBM docking station with a designed in tilt measured at about 3 degrees; this laptop came with and IBM now HGST soon to be WD HDD, so I suspect HGST's prohibition is not substantive :-).  Tom94022 (talk) 20:04, 7 January 2012 (UTC)


 * Vertical and stranger orientations are quite common in servers, consumer electronics and external drives, so yes I completely agree there not going to be a problem with this. Still, orientation could perhaps affect temperature distribution, (platter) vibration and bearing load (all probably more in 3.5" drives than in 2.5" drives), so I wouldn't be surprised if this resulted in some empirically measurable performances differences from the drives having a harder time locking onto a track. But without any reliable sources confirming this, it shouldn't go into the article. Perhaps we should add a short section on "Mounting" and summarize manufacturer recommendations, discuss some industry practices, vibration in storage arrays (enterprise drives compensate for this) and perhaps a historical note on stepper drives? It surely is a point that seems to be discussed quite often. —Ruud 20:42, 7 January 2012 (UTC)


 * I suppose a single subsection with a single short sentence in section 5, Performance, would be more than enough. FWIW, IMO orientation would not substantively change temperature distribution, (platter) vibration and/or bearing load sufficient to cause any measurable change in seek performance or HDD reliability.  Of the three it is true that orienting the long access of the drive with the air flow could produce a lower temperature rise but the difference in rise is small and at least some empirical studies have not shown any correlation between temperature rise and reliability.  Since gravity is reasonably consistent even stepper motor drives were relatively immune to orientation so long as you didn't change the orientation after formatting.  Open loop linear actuator drives (stepper and voice coil) couldn't be vertically mounted but most of them were so big they couldn't be so oriented.  An interesting historical footnote, IMO, not worthy of mention. Tom94022 (talk) 22:54, 9 January 2012 (UTC)

New technology 4k sectors
I am was trying to get info. on the new hard discs where sector size has increased from 512 bytes to 4k per sector. While this information is available at https://en.wikipedia.org/wiki/Disk_sector amd preferably https://en.wikipedia.org/wiki/Advanced_Format somehow mention of this new technology is not there in the entire article. At the very least this should be mentioned giving link preferably to the wiki page on Disk_Sector. Shirishag75 (talk) 03:01, 10 January 2012 (UTC)
 * Please see 3.2 HDD formatting wherein it states, "Modern HDDs, such as SAS and SATA drives, appear at their interfaces as a contiguous set of logical blocks; typically 512 bytes long but the industry is in the process of changing to 4,096 byte logical blocks; see Advanced Format." I suggest that is enough for this article.  Tom94022 (talk) 06:24, 10 January 2012 (UTC)

Recertification
- relocated this section from first to appropriate time sequence on this talk page Tom94022 (talk) 17:21, 17 January 2012 (UTC) I'd like to see something in the article about recertification. Recently, previously used (presumably) hard drives are being sold as "recertified". Questions to answer include things like:


 * Is there a standard meaning or process of recertification?
 * What should the user expect as to lifetime or reliability of a recertified drive?

Rhkramer (talk) 14:29, 17 January 2012 (UTC)


 * I don't see a need for such a section and unless there is a reliable source it would be impermissible OR any how. FWIW, based upon my understanding of the industry:
 * Is there a standard meaning or process of recertification?
 * No. It depends upon the vendor of the recertified drive.
 * What should the user expect as to lifetime or reliability of a recertified drive?
 * Disk drives are pretty reliable so you can expect the drive to outlive the warranty provided by the vendor but by how much is not predictable.
 * I wouldn't have too much problem buying a reconditioned or refurbished drive from an HDD vendor but I wouldn't trust anyone else. Tom94022 (talk) 17:21, 17 January 2012 (UTC)

Form Factor Section
Each form factor has multiple z-heights, therefore, I see no reason to single out 5&frac14;-inch Half Height any different than 3&frac12;-inch or 2&frac12;-inch and I reverted change. Tom94022 (talk) 17:19, 23 April 2012 (UTC)

about computer
first hi.

ineed to get information about the mounting computer — Preceding unsigned comment added by 196.201.207.225 (talk) 06:17, 5 June 2012 (UTC)

Cache
On board controller cache is a common technique to improve performance. Complex algorithms may get used or simple read-ahead. Can this be addressed under its own heading? It is a glaring omission in an otherwise excellent article.

Thanks, BAC — Preceding unsigned comment added by 124.183.125.47 (talk) 21:11, 29 July 2012 (UTC)

Too short lede?
Thumperward has tagged the lede as too short without providing any insite as to why it appears to him to be so. I for one think we have a Goldilocks lede, not too long, not to short; so I propose we give this tag a two week lifespan. If no one makes substantive changes to the lede in two weeks we pull the tag Tom94022 (talk) 00:12, 2 October 2012 (UTC)


 * I truly do not know where this misconception that the lead section is supposed to be an exercise in brevity comes from. The lead to this article is wholly inadequate: the article is 100k long while the lead consists of five short sentences. Instead of providing a summary of all of the key information in the article, it provides only the barest context (one sentence for the entirety of the discussion on design and construction; two sentences on history containing only introduction and adoption dates, and a mealy comment about advances in technology; nothing whatsoever on capacity, form factors, performance, interfaces, integrity, sales or cultural impact, which are all top-level sections in the article body). Instead of imposing deadlines on other editors, it would befit you to actually take the time to read and understand WP:LEAD. I'm going to try to work on this myself, but it would be completely inappropriate simply to de-tag it at present. Chris Cunningham (user:thumperward) (talk) 08:24, 2 October 2012 (UTC)


 * Insults are really inappropriate: I have read and do understand WP:LEAD and consider this lede at two paragraphs sufficient against a suggested "normally ... no more than four paragraphs." Most of what u want to add is IMHO detail covered by the two paragraphs and is not appropriate for a lede, but that is really your choice - go ahead and be bold (but keep it less than 5 paragraphs).  How long do you think it appropriate to keep a tag that causes no action?  I propose two weeks and would be interested to see if editors interested in this page can come to a consensus.  Tom94022 (talk) 15:40, 2 October 2012 (UTC)
 * BTW, Hard_disk_drive has been viewed 252595 times in the last 30 days. This article ranked 1197 in traffic on en.wikipedia.org. So I think 125,000 views without action might say no interest :-) Tom94022 (talk) 15:43, 2 October 2012 (UTC)


 * It doesn't matter how many people have visited the page: I'm sure we get millions of hits a month on terrible articles on Digimon characters, but that doesn't mean we should emulate them. This article will never pass GA (let along FA) without addressing this issue, and that should be the future goal. As I said, I'm planning on working on this myself, but it is completely inappropriate to hobble the efforts of others by imposing arbitrary deadlines on needed work which is tracked by cleanup tags. Chris Cunningham (user:thumperward) (talk) 17:56, 2 October 2012 (UTC)


 * You are entitled to your opinion, as indeed am I. If you don't think page view is a useful measure how about the over 300 editors watching this page who within a short time all should know about the tag.  How long do you think it appropriate to keep a tag that causes no action? Tom94022 (talk) 21:44, 2 October 2012 (UTC)


 * Until the issue is resolved. The cleanup tag is there to draw attention to the problem (particularly in this case where there is such a widespread misunderstanding of the purpose of the lead section: it's one of the most endemic problems in large articles). As for 300 watchers, that's no more relevant a statistic if those watchers aren't paying attention, and there's significant evidence that they aren't (there have only been ~500 revisions of the article in the last year, of which roughly 70% have been either vandalism or anti-vandalism, and there have only been 75 revisions of the talk page in that time as well). This is plainly not an article which gets a lot of peer review at the moment regardless of how many page impressions or watchlisters it has, and flagging it for cleanup is one tool to remedy that. Chris Cunningham (user:thumperward) (talk) 00:24, 3 October 2012 (UTC)


 * I think a lede that would be "long enough" to satisfy that rather arbitrary guideline would be as long as a short WP article all by itself.
 * Perhaps part of the answer is to split the article. I'm thinking of the average WP reader who comes here looking for information on current or recent hard drives. They're probably not much interested in the IBM 350 RAMAC, eight-inch form factor drives, or how we used to choose the optimum interleave factor, or in the changes from ST506 to ESDI interfaces. They are probably not even interested in polycrystalline recording material. Heck, the article uses the floppy disc drive form factor as a point of comparison and most computers have not come with floppy drives for almost ten years now. I'd suggest splitting off a lot of the historical and current, but deep, technical information into a "Hard disk drive history and technology" article and leave this one to describe the functional characteristics of modern and very recent devices. I'd think the typical reader could then find their desired info much more quickly here. Jeh (talk) 00:41, 3 October 2012 (UTC)


 * I agree the article is too long; perhaps one small step is to get rid of all tags that result in no action, starting with this one :-) Tom94022 (talk) 17:30, 3 October 2012 (UTC)


 * How does "get rid of all tag that result in no action" result in any improvement at all? If you have a personal dislike for cleanup tags then you can hide them using your user CSS. it is distasteful to suggest that the very thing which is a first attempt at resolving the continuing problems with this article is in fact causing them. Chris Cunningham (user:thumperward) (talk) 08:25, 4 October 2012 (UTC)


 * Right now u are the only person to express an opinion that the lede is too short. The most recent edit, in spite of the tag, made the lede shorter.  To answer your question, removing tags that have had no action cleans up the article by removing material proven by lack of action to be not necessary.  Now how about answering my question, how long to you think a tag should stay up with no action - particularly in this case where there is a genuine dispute that the tag is applicable? Tom94022 (talk) 15:29, 4 October 2012 (UTC)


 * As I said, until the issue is resolved. As for this being a "dispute", I don't consider one cherry-picked line about the number of paragraphs the lead should contain (along with lots of fluff about page views and watchlist counts which is neither here nor there) to be an adequate rebuttal of the points made (to wit, that the present lead plainly fails to serve as an adequate summary of the article as even the most key factors are given a sentence at best and roughly 80% of the article has no representation in the lead at all). As for the most recent edit making the lead shorter, that only serves to illustrate that you don't understand what WP:LEAD is trying to put across: it is not a matter of raw character, word or paragraph count, but of coverage. Tightening the wording does not act in contradiction of the cleanup tag, and indeed it is better that the lead is not padded out with fluff in lieu of proper coverage. Chris Cunningham (user:thumperward) (talk) 09:05, 5 October 2012 (UTC)

This edit, yet again, appears to place brevity above comprehensive coverage. Aside from reintroducing a whole bunch of run-on links in the first sentence, it excises material on the history of adoption of HDDs along with the important distinction between binary and decimal notation in measurement and mushes the majority of the rest of the considerations from the capacity section into he meaningless phrase "system usage and reporting". The new lead certainly wasn't too long for an article of this length, so the removal of ~1/3rd of it was unnecessary. Chris Cunningham (user:thumperward) (talk) 20:25, 8 October 2012 (UTC)


 * "The lead section should briefly summarize the most important points covered in an article in such a way that it can stand on its own as a concise version of the article." The lede sentence partially quoted above is,
 * "However, due to system usage and reporting not all of the specified storage space is ever available for user storage."
 * To paraphrase you, it provides a summary of all of the key information in Capacity section. The Binary Prefix wars are of little interest to the general user and if one wants to learn more about why his/her reported capacity differs from the specified capacity that is what the Capacity section is for.  Tom94022 (talk) 16:51, 9 October 2012 (UTC)


 * Again, "summary" does not mean "bare minimum". This in particular means that detail should not be replaced with vague hand-waving. Nor is capacity the only thing that was excised: ~30% of the new material was removed. I'm going to replace the fuller summary unless there is a detailed rationale for reducing it; these cherry-picked fragments from the guidelines certainly aren't convincing, especially considering the continued lack of demonstration of understanding the guideline (to wit, the edit summary for this response). Chris Cunningham (user:thumperward) (talk) 21:09, 9 October 2012 (UTC)


 * Please note the guideline says briefly summarize. Yr ~30% statistic is wrong (more like 26%)but also pretty much meaningless so merely reinstating yr material might look like the start of an edit war, particularly since you continue to attack my motives using pejoratives.  Most of what I did was simply move things around and simplify.  For example the lead sentence you don't like, consolidates the original first sentence and one from your second paragraph.  Likewise I changed your
 * "Hard disk sizes are typically specified in multiples of 1000 megabytes (MB);"
 * to
 * "HDD sizes are specified in multiples of 1000 - "
 * which is both more accurate and shorter.
 * The one area I made a major cut was indeed in the area of Capacity and you have provided no reason why this one section needs to be developed in detail in the lede. You also should note that I increased the size of the performance section, fixing a major oversight on your part.  So again, please stop the Ad hominems and focus on making the lede briefly summarize the article.  Tom94022 (talk) 21:40, 9 October 2012 (UTC)


 * I guess I'm just disappointed by having to repeatedly face gainsaying like "this is a Goldilocks lede" which puts subjective opinion over discussion. But let's look at exactly what I'd like changed:
 * 1. The alterations to the introcution of adjectives in the first and second paragraphs should be undone. This is a classic error in writing lead sections: the insistence in cramming every possible adjective into the first sentence. Links should not be bunched-up next to each other, and long strings of comma-separated adjectives are awkward in any case.
 * We disagree, I did take out digital as a concession. Right now there are only two commas so I don't think yr description is at all accurate. The problem I have with yr original construction is yr two sentences relate to the same material but are in two paragraphs and somewhat redundant.  Not appropriate for a lede IMO Tom94022 (talk) 16:18, 10 October 2012 (UTC)
 * 2. Mention of data loss should be restored, as data integrity is a top-level section in the article body.
 * Head crash is by most accounts one of the lesser causes of data loss in HDDs so highlighting it in the lede is not at all a summary. All devices can loose data so I am not sure why it should be in the lede at all. Tom94022 (talk) 16:18, 10 October 2012 (UTC)
 * 3. The contrast between binary and decimal multipliers should be restored. We devote a considerable portion of the relevant section in the body to that distinction.
 * But it is only one of the several causes of the reason why the reported capacity is always less than the specified capacity. The Binary Prefix wars still being waged at Wikipedia is the only reason we devote so much to it in this article.  The current sentence summarizes all of the factors into "system usage and reporting."   If you don't like that phrase why not come up with a better summary phrase instead of bring the entire section into the lede?
 * 4. I'm not sure it's accurate to describe server hardware as using 3.5" disks. If we're going to highlight server disks, it should be to point out that they use FC / SCSI / SAS and are predominantly hotpluggable rather than to focus on the form factor (which varies considerably). This needs expanded in the body as well, to be fair.
 * That's not what it says, all it says is desktops and servers typically user 3.5" form factors which is generally true although some do use 2.5" form factors. FC and parallel SCSI are obsolescent perhaps obsolete so don't belong in the lede any more so than ATA and ST412. Tom94022 (talk) 16:18, 10 October 2012 (UTC)
 * Chris Cunningham (user:thumperward) (talk) 09:20, 10 October 2012 (UTC)
 * My responses embedded above. To make it work I had to edit your numbering system, hope this doesn't cause a problem for you. Tom94022 (talk) 16:18, 10 October 2012 (UTC)

replying:

Chris Cunningham (user:thumperward) (talk) 08:57, 11 October 2012 (UTC)
 * 1) "Only two commas" is hardly an excuse. No article should ever have four juxtaposed adjectives, all linked, in its lead sentence. It is, as I say, a very common error in writing articles here whereby editors believe that it is absolutely necessary to summarise the entire article in its first sentence. It is not. To pick a random example, Barack Obama's initial sentence says little more than that he is the current POTUS. You could cram a dozen very important adjectives into that sentence, but it wouldn't improve it.
 * 2) On further investigation, you're right that data loss is not given so much coverage in the article body that it needs included in the lead. Happy with that removal.
 * 3) I think that the previous summary, "Not all of the disk may be available for storage, as the system may contain inbuilt redundancy for error correction and recovery and the file system used will reserve some disk space for its own use", was concise enough for the lead and yet not as mushy as "However, due to system usage and reporting not all of the specified storage is ever available for user storage". For the binary prefix thing, the fact of the matter is that we devote considerable space to it in the article body and so it belongs in the lead. One solution may be simply cutting down on how much time we give it in the body.
 * 4) I think "obsolescent" is a rather strong way of putting it. PCs are no longer sold with PATA drives, but "obsolescent" implies a rather stronger deprecation. The vast majority of present hardware still uses these technologies, which isn't the case for the likes of ST412. The point here, though, is that server hardware shares little in common with desktop hardware, and the phrasing used somewhat implied that servers used basically the same disks as desktop machines (which they don't).

replying:

1. We disagree, but how about:
 * A hard disk drive (HDD; also hard drive, hard disk, or disk drive) is a magnetic data storage device featuring low cost, non-volatility and random-access. [with appropriate wiki links]?
 * BTW, while you clearly have strong opinions about adjectives in sentences, I can find nothing to support your opinion. A search of style guides i can find only gives guidelines on how to use commas with multiple adjectives but say nothing about limiting there use to any number much less your apparently arbitrary statement that four is too many.  The focus is on clarity, and right now the four seem clear enuf to me.

2. Thank you

3. We disagree, but how about:
 * However, not all of the specified storage is ever available for user storage because of file system usage and possible redundancy, amongst other things.
 * Frankly I think "system usage and reporting" pretty much covers all of the non user consumption of disk storage and is not at all mushy. I note you have already shrunk the Binary Prefix section, have at it.

4. I don't feel so strongly about PATA and FC, but I would have u note:
 * "The transition from IDE to SATA drives is almost complete. SATA holds more than 90% of market share. (Sep 2009)"
 * "SAS will account for 57% and SATA for 43% of the total multiuser HDD market by 2013, ..." (Gartner 2009) That is FC will amount to 0%.
 * I think that after 4 years with little market share PATA is obsolescent and is approaching obsolete (one of the reasons I finally gave up my IBM T23 laptop is I couldn't get a big enough 2½-inch PATA upgrade, having started with 60 GB and going thru several upgrades to the maximum at 250 GB). FC is probably a year or two behind.  Since this is the lede I don't think we should waste the space on these interfaces.  I'm not sure what in the phrasing implies "servers used basically the same disks as desktop machines" - they do share the same form factor and as the statistic above indicates there are a lot of SATA drives in the server market today.  Perhaps a sentence in the lede summarizing the market segmentation (Section 9), although to be honest I really don't like adding to the lede?

Tom94022 (talk) 16:40, 12 October 2012 (UTC) [added later]
 * This is a subjective style matter, but I have little doubt that it would be brought up in a peer review. It's not obvious why you are insistent on retaining it. I haven't undone it pending discussion, but I will do if there's no better reason than "clarity" (frankly, I think jamming groups of jargon words next to each other reduces clarity for anyone except a non-expert, and that appeasing experts is the main goal of insisting that the first sentence be so comprehensive).
 * Glad we've come to an agreement here.
 * I don't understand what the removal of specifics in favour of mushier generic terms gains us here. Can you explain why you're opposed to a revision which is only slightly longer but much more accurate?
 * This isn't an article exclusively for Western owners of modern desktop computers. It should not be presented from the point of view of the bleeding-edge consumer. It should certainly explain what the "current" technologies are, but it would be misleading to suggest that the technologies in use on the majority of extant consumer hardware (which is over five years old) are somehow historical relics. That's even more relevant when it comes to server hardware, which has a far slower lifecycle.
 * (Addressing market segments): the article as a whole is badly skewed in favour of technical content over real-world impact. The current "market segments" section needs significant expansion. Once that's done, it deserves more attention in the lead (which still has plenty of potential for expansion).
 * Chris Cunningham (user:thumperward) (talk) 10:24, 12 October 2012 (UTC)


 * 1. Even thought I don't agree it is necessary, I have made the change I proposed above. Now it has three nouns strung together ?-:
 * 2.
 * 3. Actually the current construction is more accurate and not at all mushy since it includes all sources of loss of storage in words that are understandable to all using their common dictionary meaning. The proposed construct is wordy and only addresses two of the several, but if you insist on this construct here is a less wordy version of your proposed language:
 * "However not all of the specified storage is available to the user due to amongst other things, file system usage, how the system reports capacity and possible redundant usage for error recovery."
 * I really think this adds nothing but a lot of words that are accurately covered by "system usage and reporting." You have characterized this as "mushy" - I don't think it is mushy at all:
 * system usage covers hidden partitions, RAID, file system and any other usage by the system
 * and [system] reporting covers binary reporting, truncation, and any other reporting difference by the system
 * so I would really appreciate what you find mushy about this simple and accurate construction.
 * 4. I haven't seen any statistics but I would bet given the retirement rate of old computers PATA is no longer a large share of the market. And of course FC was hardly measurable. I agree that the lede should not be aimed at bleeding edge western consumers and I don't get your point that leaving out PATA and FC somehow makes it so.  I thought the lede was for the general audience in which case even SAS would be in appropriate.  Be that as it may, if you want to add PATA and FC to the list go ahead.  I won't even argue about long lists of nouns being inappropriate
 * 5. I don't follow your line of reasoning that correlates length of article to length of lede and here I don't follow the similar line that a short section doesn't belong in the lede. You will probably accuse me of cherry picking but, "The emphasis given to material in the lead should roughly reflect its importance to the topic" - not its length in the article!
 * Tom94022 (talk) 16:40, 12 October 2012 (UTC)

The new lede is poorly written. The purpose of a lede, to my mind, is to give enough context that the user can navigate the rest of the article. If a given fact is so detailed that it provides no additional context outside of itself, it shouldn't be in the lede. A few specific critiques: That's all in addition to grammatical and phrasing errors. (e.g., last sentence of first paragraph: cost per unit of reliability?) The previous lede, however brief, was better than this one. Here's how I would set it up: History, interfaces, and a detailed discussion of the mechanics of operation are better left to the main article. Alexdi (talk) 19:52, 16 October 2012 (UTC)
 * Too many vague descriptors (e.g., "thin layer," "tiny gaps," "rapidly rotating")
 * Nonsensical and tangential discussion of capacity (If units of 1000, why 1024? Why care?)
 * Performance section provides metrics with no context. Why would a reader care about this? What's important is to get a feel for where HDDs stand in the hierarchy of storage caching.
 * Size section presumes a knowledge of 'form factors'. What's 2.5"? The width axis?
 * Why are revenues listed instead of units shipped or market penetration? Given the massive price increases earlier this year, high revenue figures probably present an erroneous picture of growth.
 * General classification and purpose
 * Speed and capacity in caching hierarchy; reference SSDs and RAM for context; three relevant metrics: capacity, random access speed, sustained throughput
 * Physical size and weight
 * Power use, also as compared to SSDs; relevant metrics: idle vs. active seek
 * Reliability, including HDD-specific weaknesses (e.g., impacts)
 * Market penetration and trends

SSD replacing HDD?
This is certainly POV since most neutral observers state that SSDs are not likely to replace HDDs in general purpose computer systems in the foreseeable future. Furthermore, the cites really don't support the assertion that SSD are more reliable. Finally, since there is no such discussion of SSDs replacing HDDs in the body of the article, it is inappropriate in the lede. So if you want to put something into the body, that's great, but even then I would argue that it is not sufficiently important at this time to add it to the lede. Tom94022 (talk) 16:23, 9 October 2012 (UTC)


 * Who are these "neutral observers"? *All* the tech magazines and web sites are talking about SSDs replacing Hard Disks. How can it be POV if it has 3 references? The problem in the lead is that it says HDDs have maintained their dominant position (against what??) when in fact SSDs are way faster and possibly more reliable. HDD vs SSD is one of the biggest stories in computing at the moment. Why ignore it? Bhny (talk) 16:34, 9 October 2012 (UTC)


 * Maybe you didn't read my references. They are all recent and all talk about how SSDs are faster, more "resilient" and are replacing HDDs in desktops and servers. Obviously hdds have lost out to solid state in laptops and other mobile devices but for some reason you think a laptop isn't a general computing device. Bhny (talk) 17:42, 9 October 2012 (UTC)


 * Again who is this neutral observer? Bhny (talk) 17:43, 9 October 2012 (UTC)


 * Anecdotes are not evidence of change in a dominant market position nor is it obvious that SSDs have made a substantial inroad into the laptop market. Yes laptops are in the gp market segment and a few are sold with SSDs but way more than the majority of laptops are sold with HDDs, see, e.g. "Hard drives still dominate SSDs in laptop market".  And the penetration of the mainframe, server and desktop market segments is miniscule.
 * The neutral observers are the independent market analysts like IDC, Gartner, Needham, iSuppli, etc for both current market sizing and projections to the future; none of them have HDDs loosing dominance in the foreseeable future. The bottom line is that SSDs will not close the price/terabyte gap of HDDs in the foreseeable future and therefore will be limited to early adopters.
 * The reason the gap won't close is that both are now proceeding down similar areal density learning curves. Flash had an advantage early in this century because it piggy backed on Moore's law (then driven by DRAM) but now it is the bleeding edge so it won't accelerate.  U must know that the semiconductor folks are projecting Moore's law is slowing down. Furthermore, the demand in hand held device market is so large that there is right now, little incentive for the Flash manufacturers to lower their chip price to close the gap
 * The analogy I would give you is hybrid versus conventional engine cars - better mileage, lower maintenance, faster off the line, but not likely to dominate the market until when and if the cost gets close. Tom94022 (talk) 19:40, 9 October 2012 (UTC)


 * You might take a look at iSuppli SSD & HDD Market Tracker which shows an 8% current market share on a unit basis for SSDs. The share is projected to grow to about 22% by 2015 but the unit growth is almost the same for both products.  Sort of like when Russia was outgrowing the US and u know how that turned out.  Even if the relative share growth does occur, HDDs with 78% market share could be reasonably called dominant even then.  If you measure share on a storage basis HDD is even more dominant.  I don't dispute that in some applications some HDDs are being replaced by SSDs but there is simply no evidence that HDD dominance of the market is going to be challenged in the foreseeable future.  Tom94022 (talk) 00:00, 10 October 2012 (UTC)


 * I would also suggest that while the SSD does have a foot in the door (but only that) as the primary drive in a PC, when people add more storage it's almost always in HDDs. For the home market, look at the large number of models of external HDs. And in large server storage, hard drives are even less challenged. Jeh (talk) 06:10, 10 October 2012 (UTC)


 * Given these considerations and Tom94022's refs, I am removing the "until recently" qualification from the lede. Jeh (talk) 06:10, 10 October 2012 (UTC)


 * ok, new edit where hdds are still dominant but for reasons of cost and capacity. I can't see any argument against this. Bhny (talk) 08:40, 10 October 2012 (UTC)


 * I made the change before reading this dialog; HDDs are continuing to improve in both reliability and speed so it is not just cost/storage. Tom94022 (talk) 18:11, 10 October 2012 (UTC)


 * We are getting closer. Of course they have improved in speed but that isn't how the sentence parses- it seems to say they remain dominant because of speed etc. while they are actually losing share because they are much slower than SSD Bhny (talk) 18:19, 10 October 2012 (UTC)
 * I agree. HDDs may be getting faster, but this is not why they're holding their ground against SSDs. – Steel 18:41, 10 October 2012 (UTC)


 * My problem with the sentence is that to me it implies that progress has only being made in cost/storage and reliability while in fact HDD have continuously improved in all of their measurements since the beginning. Take a look at IBM's first HDD versus its last HDDs for one set of measurements for the first 46 years - today's drives are probably a factor of ten better than that.  I think the idea of 60+ years of continuous improvement (longer than Moore's law) is an important point for the lede and somehow it gets lost in this focus on competition with SSDs.  Any ideas?  Tom94022 (talk) 00:51, 11 October 2012 (UTC)
 * Maybe the SSD thing should be moved to the end of the lead instead of being mixed up with this point about the historical dominance of HDDs. The rise of SSDs is important but I'm not sure it belongs in the very first pargraph. Tbh I think that would probably round off the lead quite nicely: a few paragraphs about HDDs, their history and mechanics, and then a final one about future issues, challenges, and the role of HDDs at present and in the future. – Steel 13:19, 12 October 2012 (UTC)

I've moved and reworded it. Chris Cunningham (user:thumperward) (talk) 11:35, 19 October 2012 (UTC)

File:Seagate ST33232A hard disk inner view.jpg to appear as POTD soon
Hello! This is a note to let the editors of this article know that File:Seagate ST33232A hard disk inner view.jpg will be appearing as picture of the day on October 23, 2012. You can view and edit the POTD blurb at Template:POTD/2012-10-23. If this article needs any attention or maintenance, it would be preferable if that could be done before its appearance on the Main Page so Wikipedia doesn't look bad. :) Thanks! — howcheng  {chat} 19:09, 22 October 2012 (UTC)

Dubious
Image caption tagged dubious because the image *itself* has been tagged. (Note that I only recently placed a tag on the SVG image, but as it was originally just a vectorised version of a bitmap image- i.e. based on the same data- and *that* had been tagged since 2009, the issues remain).

See File:Hard drive capacity over time.png (the original) and File:Hard drive capacity over time.svg (vectorised version, was later updated, but only for new data, existing data still has same problems).

Ubcule (talk) 14:28, 16 June 2012 (UTC)


 * Since it is dubious I propose taking it off this article. Note I am the person who marked the image source as dubious.  Tom94022 (talk) 18:11, 16 June 2012 (UTC)


 * So, your concern is, as you write at the image's page: at least in the 1980s misses the higher capacity drives which generally were not offered at retail


 * So what? "not offered at retail" means they were not on sale. Sure, there are always better processors and hard drives and whatnot, but if one can't purchase those, then those shouldn't "generally" be noticed at all. 85.217.14.135 (talk) 21:25, 18 August 2012 (UTC)


 * Perhaps I should have said, not advertised at retail. High capacity disk drives were for sale and in the early 1980s were 70% or so of the market as measured by revenue. High capacity drives were not advertised for sale thru retail channels such as stores and advertisements in magazines (e.g. Byte).  The high capacity disk drives were generally sold by the direct sales forces of computer manufactures (e.g., IBM, CDC, DEC, etc.,) and/or their resellers (e.g. Systems Industries).  IBMs prices were published and rarely discounted but the others were frequently discounted so calculating an average price is difficult.  The higher capacity market segment was by far the largest market segment by revenue for most of the 1980s making this graph of retail drive prices pretty much meaningless for at least the first half of the 1980s.  Tom94022 (talk) 21:51, 18 August 2012 (UTC)


 * And you claim 2520 MB disks were available. Maybe, but they must have been extremely rare. As I recall, I got a new computer with a 60 MHz Pentium and 400 MB hard drive. And that was over ten years later! 82.141.93.49 (talk) 10:51, 19 August 2012 (UTC)


 * There was a world before the PC, see for example IBM 3380 direct access storage device. Tom94022 (talk) 07:18, 20 August 2012 (UTC)


 * Doesn't that mean they weren't sold at retail, but rather to companies who assembled the whole computers? And what is the source for that 70% share? 82.141.64.59 (talk) 04:30, 23 October 2012 (UTC)


 * Retail is the sale of goods and services from ... businesses to the end-user. Some mainframe companies (e.g. IBM, CDC) and some minicomputer companies (e.g. DEC) built there own disk drives while others (e.g. Data General) purchased them from third parties. There were also system integrators who purchased HDDs.  Regardless they then sold them to the retail (end-user) customer thru their direct sales forces. What we didn't have in 1980 was many retail computer shops and although some disk drives were sold by mail order thru advertisements in publications such as Byte, they were very few and collectively and not representative of the market. The source for market share is 1981 Disk/Trend report.  Tom94022 (talk) 22:13, 23 October 2012 (UTC)

Capacity Fudge
The fudge is not necessary. There is ample evidence on the web that the capacity reported by the system is always less than that advertised by the manufacturer for at least one of the reasons selected. Google "Hard Drive Capacity Always Less Than Stated" and read the several reliable sources identified. The problem is that there is no one concise source that lists all of the factors the way we do in this article. I am not aware of any evidence to support a may be fudge and so am reverting the change. Tom94022 (talk) 19:45, 1 November 2012 (UTC)

Platters column in Form Factor
The Platters column in Form Factors always causes confusion since it is not always the highest capacity drive that has the maximum number of platters per drive and historically there may have been earlier models with many more platters than the current highest capacity HDD. For example, today in 3½-inch the highest density is about 1 TB/platter but there are many 5 platter drives currently around at less than or equal to the current maximum 4 TB/drive. Historically, I think there have been as many as 10 platters in a 1.63 inch high 3½-inch FF. WD (HGST) has just announced a helium filled drive for next year that will allow 7 platters in a 25.4 mm high 3½-inch FF. I think we should just eliminate the column as TMI and confusing at that. Perhaps include it as text in the details on the FF in the section. Any better suggestions? Tom94022 (talk) 18:15, 15 November 2012 (UTC)


 * I'm also concerned with the title of the table. It appears to be PC based and to only cover recent drives. Older drives such as the IBM 355 and the IBM 1301 had far more platters than anything in the table, as did disks from other vendors in the first third of the 20th century. Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:50, 20 November 2012 (UTC)


 * The term form factor generally has not been used to cover 24-inch and 14-inch disk drives although I suppose one might consider some 14-inch drives to be in an ASME rack form factor - so I don't have any real problem with the table title. The section does cover the shapes of the early drives you mention (washing machine or rack).  Adding more about these historical drives would IMO be TMI.  The term form factor didn't come into general usage until the early 1980s when first the SA1000 HDD was the same form factor as the FH 8-inch FDD.  This also is covered in the section.  The form factors listed are not in any way limited to PC since enterprise HDDs comport to the listed form factors.  I'm not sure what you are proposing, expand the chart to include the early models - I would oppose that as TMI.  Tom94022 (talk) 23:23, 20 November 2012 (UTC)

Unknown answer while reading the article
I have question about hard drives and I didn't find the answer in both French and English version of the article. I am wondering about the technical reason that prevent hard drives to have two or more heads stacks as shown by this example: In this case it could increase random access possibilities, and thus increase the speed at the expanse of less capacity. A typical usage would be fetching data over a fragmented partition or copy a file to an another place on the disk. It could also be a security in a case a head stack got a failure without damaging any platter.

But it is not possible because it is so much evident. I couldn't find the answer anywhere.English Wikipedia could give more technical details about hard disk so it would be useful. — Preceding unsigned comment added by Ytrezq (talk • contribs) 18:02, 3 February 2013 (UTC)


 * There is no technical reason to not have multiple head stacks for one set of disks and there have been a number of such drives over time starting with variants on the original RAMAC which I seem to recall had provision for up to 8 head stacks and shipped at least one variant with two or perhaps even four head stacks. Making the head stacks read the same tracks is a challenge, there have been many drives with two head stacks on one set of disks, e.g. the IBM 3380, but the last drive I know of that could read the same track with heads from two head stacks was a Conner product circa 1990 which had only limited production.   The market seems to have decided this doesn't make economic sense. On the whole, I think this is not notable so it shouldn't be in Wikipedia.  Tom94022 (talk) 21:23, 3 February 2013 (UTC)

HDD projected to remain more popular "for the foreseeable future"
The reference for this is from someone at Western Digital. This is like someone at Kodak talking about how film will never replace digital. I've balanced this dubious opinion with a reference to a Time.com tech article. It is POV pushing to only have one side on this. Someone strangely reverted my addition. I've put it back and corrected a link error. Bhny (talk) 00:30, 25 November 2012 (UTC)


 * I reverted and again will revert your edit since the cited reference (URL now corrected) does not support your POV statement that "though cloud storage could make HDDs obsolete for the PC." The reference notes cloud storage has not been successful in music and most consumers "appear to be happier buying songs outright from Apple and being responsible for storing them on local disks. And hard drives have gotten so cheap that there’s no economic reason not to have one or more humongous ones."  Hardly a forecast of cloud storage obsoleting HDDs on the PC.


 * While the reference is authored by WD, Forbes is a reliable source and there are plenty of other reliable sources including those in the Flash and SSD industry that have made the same forecast; e.g. Eli Harai CEO SanDisk or The SSD Guy. It is possible at some time in the unforeseeable future that HDDs would be obsoleted by some other technology, possibly even Flash, but at this point it is pure speculation and inappropriate in an article, much less in the lede. If you can find a reliable source of a forecast, not anecdotes, great, but your current reference does not rise to the that level.  Tom94022 (talk) 01:09, 25 November 2012 (UTC)


 * from the article "In certain respects, it’s already happening." "The trend is even creeping over to traditional PCs." Bhny (talk) 01:29, 25 November 2012 (UTC)
 * Tom94022 is correct: In general, an article about a specific thing should not speculate about the future (even if sourced to someone's opinion). Of course, HDDs will one day be obsolete, and perhaps people will be silly enough to store their data in the cloud, but it is not the role of Wikipedia to promote inherently unprovable claims in an article like this. Johnuniq (talk) 01:31, 25 November 2012 (UTC)
 * So you are saying we should remove "for the forseeable future"? Bhny (talk) 02:05, 25 November 2012 (UTC)
 * Apparently this boils down to opinions expressed by "Currie Munce ... VP, Research & Advanced Technology, for HGST, a unit of Western Digital" here and "Harry McCracken .. editor at large at TIME, where he writes about personal technology for the magazine and TIME.com. He’s been a gadget nerd since the late 1970s and is the founder of Technologizer and the former editor in chief at PC World magazine" here.
 * Yes, a WD employee can be expected to talk up HDDs, but the company is backing that opinion with money, and the source satisfies WP:RS. By contrast, a technology editor is just expressing their view (and they have to express a view every week to fill space). Perhaps the text could be recast, but it is reasonable to give the opinion of a major disk manufacturer in the article, and it is not reasonable to "balance" that with essentially an arbitrary person's opinion. Johnuniq (talk) 05:02, 25 November 2012 (UTC)
 * If you're going to have projections, then representing a range of projections in a fairly balanced way makes sense. Why ignore the alternative POV here?  Dicklyon (talk) 05:27, 25 November 2012 (UTC)

To a certain extent I feel like we are in the midst of the argument between Nate Silver and the talking heads over the 2012 election. On one side we have executives of major companies on both sides of the technologies and based in their analyses saying SSDs will NOT replace HDDs in the foreseeable future and on the other side we have columnists enthralled by the speed of HDDs saying it may happen because that's what their gut tells them. If you carefully read the four cites by Bhny, none of them say it will happen; the best says: ""It's unclear whether SSDs will totally replace traditional spinning hard drives, especially with shared cloud storage waiting in the wings. The price of SSDs is coming down, but still not enough to totally replace the TB of data that some users have in their PCs and Macs. Cloud storage isn't free either: you'll continue to pay as long as you want personal storage on the Internet. Home NAS drives and cloud storage on the Internet will alleviate some storage concerns, but local storage won't go away until we have ubiquitous wireless Internet everywhere, including planes and out in the wilderness. Of course, by that time, there may be something better. I can't wait." [emphasis added]"

And even if local storage goes away, the stored information won't and where would it be in such a scenario, on HDDs in the cloud. Two of the cites are just articles by editors sharing their experiences and not predictive. The forth cite, discussed above, merely raises an unanswered question. While I think we make a mistake here by giving equal weight to reliable sources on one hand and POV on the other hand, in the interest of coming to a consensus I rewrote the sentence to balance the citations and distinguish between the two perspectives therein. Tom94022 (talk) 19:32, 25 November 2012 (UTC)
 * I think the reason this discussion continues to get stuck on this point is because everyone tries to make blanket statements like "SSDs are replacing HDDs" and "no, HDDs are still dominant" without regard for where and how these changes are happening, or aren't happening. No backup storage datacentre is going to be replacing their 3tb hard drives with SSDs anytime soon. But neither are iPads or Macbook Airs ever going to ship with rotating hard drives. I have made an edit to this effect which should satisfy everyone. – Steel 20:51, 25 November 2012 (UTC) (P.S. the Nate Silver analogy is so bad as to be actually kind of insulting)
 * Thanks for the edit, hope u agree with my changes. Sorry u didn't like the Nate Silver analogy but I think it is exactly what is going on.  Tom94022 (talk) 21:05, 25 November 2012 (UTC)
 * There's cloud storage using SSD already. Windows Azure is "all SSD" [][] Bhny (talk) 22:13, 30 November 2012 (UTC)
 * Anecdotes are not reliable sources, particularly when they are third party rumors. FWIW, I can find nothing from Microsoft confirming Azure is all-SSD and there is some anecdotal evidence that it is not true Even if true, such anecdotes are not sufficient to establish a reliable source for a forecast of a major technology change. Remember when optical was going to replace disk or tape, take your choice.  There was some anecdotal evidence from some early adopters and advocates but it never happened.  Already the anecdotes are forecasting the demise of Flash as it scales down, e.g. Flash is dead ... but where are the tiers?. Tom94022 (talk) 19:14, 3 December 2012 (UTC)
 * That Register article was pretty bad ha. Flash is going to be replaced by imaginary things we don't know about. Anyway whether Azure is all or part, SSD is already replacing HD in the datacenter, contradicting what User:Steel says above Bhny (talk) 00:53, 4 December 2012 (UTC)
 * No one is denying that there is some penetration of SDD into the Enterprise secondary storage market, but the issue is dominance ("more popular in the foreseeable future") not anecdotal evidence of some penetration. The Enterprise HDD Market is about 10% of the HDD market (from numerous sources) or something north of 50 million units per year and not projected to change much in the next several years.  A recent market forecast by Objective Analysis a reliable market research firm projects the 2016 unit forecast for SSDs in the enterprise at 3.9 million units or less than 8% of the market measured in units four years from now.  Way less when measured in TB.  This is consistent with the forecasts of all the market research firms.  No matter how you cut it there is no reliable source for any forecast that SSDs will become more popular than HDDs in any storage market segment in the foreseeable future.  Tom94022 (talk) 04:46, 8 December 2012 (UTC)
 * No one is denying... User:Steel was denying it above Bhny (talk) 04:53, 8 December 2012 (UTC)
 * If you read my post, I deliberately specified a backup storage datacentre as a place where SSDs aren't making inroads, since cost and capacity are everything and the extra milliseconds of latency are irrelevant. This is different from, say, platform as a service infastructure which can and does benefit from SSDs because I/O latency is important. As I said before, this dispute would go away if we stopped making broad claims like "SSD is already replacing HD in the datacenter" without attention to where and why. – Steel 15:46, 8 December 2012 (UTC)

Edit war prevention :-)
I just had a "near" vs "foreseeable" (future) single-round flip-flop with Tom94022. I didn't want to get anybody's back up by summarily undoing the undo. It was just easier to flop back the flip. "Near" is better though. We don't need to quote refs verbatim, we are allowed to improve on their words. The thing about "forseeable" is that all it takes is one "existence proof" to invalidate. That is, if one person can "foresee" a future, however far out, where SSDs are superior in every way (including costs) for either medium or secondary, then that future IS foreseeable. It's a hair-splitting technicality I know, but "near" is clearer for the encyclopedic form that we are writing. "Foreseeable" is okay for a magazine form and for a writer under a deadline. But, we have the time to improve on it and a mandate to write in a different way.108.7.163.236 (talk) 19:57, 12 February 2013 (UTC)
 * Actually "one person" is not the criteria for inclusion, a reliable source is and so far there is no reliable source predicting the replacement of HDDs by SSDs as the dominant secondary storage at any time in the future. Near on the other hand seems to imply that things might be different at some unspecified time beyond the near future.  Thus it appears to me you have changed the meaning of the reference and have now imposed your own POV into the article.  Since you don't seem to like forseeable, to me a perfectly acceptable term I will now change the text to again reflect the reference.  Tom94022 (talk) 07:31, 13 February 2013 (UTC)
 * Once Wikipedia re-invents itself to be the premier source for market forecasts, it will be appropriate to include surveys of market forecasts. Until we reach that point, I suggest we stick to the observable universe as our proper focus. When the topic becomes the Cloud and the datacenter we have shifted modes from the informative to mere marketing. And, disk drives don't need a POV. Chasmo (talk) 04:26, 14 February 2013 (UTC)

Power use
It is very difficult to find the wattage of a hard disk drive. This article does not seem to discuss the topic either, at all, even though it is rather broad. 83.71.103.45 (talk) 05:16, 28 February 2013 (UTC)


 * See for example this data sheet for Seagate's current desktop hard drives. Similar information is available on every HD maker's web site. Re the article, WP is not intended to copy product information at the level expected of manufacturers' spec sheets. (Wikipedia is not a parts catalog.) It might be feasible to include typical power consumption figures for the various common current form factors, but I don't know what we'd use for references or how we'd define "typical." Jeh (talk) 06:25, 28 February 2013 (UTC)


 * It's probably worth making "power" a subsection in "Performance characteristics" given that is one of the key purported competitive disadvantage versus Flash. Tom94022 (talk) 16:57, 28 February 2013 (UTC)

ref name tags
I've removed a whole bunch of these, as they were causing errors due to (I believe) not actually being cited anywhere in the article. Feel free to re-add them if you can find somewhere to use them. Luke no 94 (talk) 21:36, 28 February 2013 (UTC)


 * Yes, as a result of your earlier change, they were no longer being cited in the article; they were cited only in the table that you removed. (I also noted that on your talk page.) Guy Harris (talk) 21:46, 28 February 2013 (UTC)


 * Which raises the question of whether useful information was removed by removing the table. If the same information was in the text then shouldn't the links have remained good?  Perhaps a culling of redundant info rather than cropping the table is a better approach.  I really don't have the time to look into this now, so i invite other editors to take a look. Tom94022 (talk) 00:11, 1 March 2013 (UTC)


 * I feel that that much data better appears in a table than in prose - trying to force it all into prose is, well, forced. While we're at it we could add a "typical power consumption" column. Jeh (talk) 00:44, 1 March 2013 (UTC)
 * Ah, it was me who messed it up in the first place (oops). The data would be fine in a table, but if it is causing layout breaking issues - even on my 1920x1080 resolution screen - then it needed to go. I would have absolutely no issue with anyone readding refs somewhere useful, or readding a better thought-out table, but I couldn't get the other table to work after some playing around, so I nuked it. I'm questioning the logic of how the ref name tags are actually used in the article: they kinda seem a bit messy, hence why this issue materialized in the first place (if they were more conventional, this would never have happened!) Luke no 94  (talk) 08:11, 1 March 2013 (UTC)