Talk:Hard disk drive/Archive 1

Archive 1

Archive 2

Archive 3

→

Archive 5

Taxonomy/etymology

The differences between fixed disk and removable disk, hard/rigid disk and floppy disk should be brought out. Now they are not. There are merely a few notes. I changed redir of fixed disk here, but it would need a page of its own, for example. Not it almost reflects current usage as most fixed disks are hard disks. Not all hard disks are fixed disks, though.

There should probably be a clear tree and explanation on the terms by physical stature and function. Perhaps on mass storage or some such page. Then you could add the definitions and jumps to the different related articles.

IEC Binary Prefixes

My feeling is that all the storage capacity units in this article should be changed to the IEC Binary Prefixes for clarity. The only objection I can see to this is that the traditional usage of the metric prefixes is more widely recognized. However, I feel this is trivial since the binary abbreviations (i.e. KiB, MiB, GiB, as opposed to KB, MB, GB, respectively) are very similar to their metric counterparts, the difference will not be noticed by most readers, but it does a lot in the way of accuracy for this somewhat confusing matter. In any case, the current state of the article is inconstistant in its use of IEC binary and metric prefixes, and something should be done either way. Please post your comments either supporting or objecting to this, I didn't want to make this kind of an edit without first consulting some others. -- uberpenguin 18:10, 2005 Mar 28 (UTC)

---

Hmm, would anyone be willing to create and/or add a schematic of the interior of a hard drive?

I've taken a few photos of the interior of hard drives that I've licensed under Creative Commons. This one might be appropriate. -JS

I was wondering what typical random access times might be for various types of drives. I just ran a benchmark program on my HD and I was wondering how it compared to others. KM

Seek time needs clarification

Typical current mass-market drives are quoted as having seek times of around 9ms. this is just the time to move the head to the correct radial position, but it doesn't include rotation of the disk to the right sector. to get the real seek time, you need to add rotational latency, which is 4.1ms for a 7200 RPM drive (60 seconds per minute divided by 7200 RPM divided by 2 (for average)) and you have an average access time of about 13.5ms. A seriously high-perforance drive has a seek time of under 4ms and latency of 2ms - i.e., 6ms average access. But these drives sacrifice capacity to achieve their very high performance, require a SCSI interface, and are fairly expensive. Tannin

Other

Tannin, any particular reason for reverting my change to an auto-generated thumbnail? There's really no reason to have separate thumbnail images anymore, and that particular image has already been listed on Votes for Deletion after being replaced at Computer storage. DopefishJustin 03:44, Apr 26, 2004 (UTC)

Yes Justin. It looks terrible, and defaces a perfectly good photograph. If you want to fuzz-up an image, do it to one you took yourself. Tannin

Mr Anon, please note that a hard disk is the internal part of a hard disk drive. The terms are not synonyms.

MFM History

I have no idea why the article claimed that MFM was invented for floppy drives. As far as I've been able to determine, MFM was first used on the IBM 3330 in 1971. The 3330 was most certainly NOT a floppy drive. The first use of MFM for a floppy drive appears to have been the IBM 53FD drive, introduced in 1977. --Brouhaha 08:04, 3 Sep 2004 (UTC)

damage

There's nothing written about the damage to hard disks: we should have at least a cursory explanation of the terms MTBF (mean time before failure), bad sectors/blocks/clusters, BBR (bad block relocation), handling by software and by hardware, hot swapping, something about refurbishing... --Joy [shallot] 14:29, 27 Sep 2004 (UTC)

I *highly* second the suggestion for more info on failure modes. The article discusses head crashes in some detail, but are there other failure mechanisms? I've recently had a rash of early disk failures in a low-end server application. I'm suspicious that the cause was a substantial number of daily rsync operations that would have hit the directory tracks very hard on a daily basis. - And it was the directory tracks where the problems seemed to lie, none of the file data itself was every lost or corrupted, just the allocation table and extent maps were repeatedly corrupted. Is there a failure mechanism driven by repeated read/write cycles to the same area of the disk? (Dave Etchells, not a registered user)

Repeated use of specific sectors is not a common failure mode. The only thing similar is a particular use pattern may expose some design flaw that wasn't that obvious before. --A D Monroe III 21:39, 20 January 2006 (UTC)

A discussion of HSA latch technology should also be included. --Wikiwonki 15:15, 10 August 2006 (UTC)

Decimal vs Binary

It seems like the following claim is incorrect.

“It is important to note that hard drive manufacturers often use the decimal definition of a gigabyte or megabyte. As a result, after the drive is installed it appears that a few gigabytes or megabytes have disappeared. In reality computers operate based upon the binary numeral system. In the decimal number system a gigabyte is 7.5% smaller than in the binary number system. The term "1.44 MB", often used to describe 1440 KB floppies (actually 1.47 MB or 1.4 MiB), introduced an anomalous definition of "megabyte" as 1 × 103 × 210 bytes (1 KKiB).”

The manufacturers actually state a smaller raw capacity of their drives. That is a 30GB harddrive is actually 30GB X 1.024.

The reason that the actual HD is smaller than the advertised number is because some bits are reserved for sector information and thing of the like depending on the OS that’s used.

This is my take on thing. Please let me know what you think, so that we can update this article if required.

No, the text in the article is correct, if a bit innaccurate on a technicality. The issue here is actually the binary PREFIX versus the metric prefix. This is largely historical; when this first became an issue there were no binary prefixes, so 2^10 bytes was called a kilobyte since 1024 (2^10) is close enough to 1000 (10^3). However, the problem becomes more noticible when you get into the multiple gigabyte range. Most OSes report a GB as what is now called a GiB, or Gibibyte; 2^30 bytes. Hard drive manufacturers continue to use the metric definition of a GB: 10^9, which is obviously less than the binary Gibibyte, because it makes it seem like the drive has a higher capacity than it actually does.

The blocбks used for partition and file system accounting information rarely occupy more than a few KiB, and is utterly insignificant when talking about discrepencies of tens of Gibibytes. I'll edit this section of the article to be a little more clear. -- uberpenguin 15:16, 2005 Apr 4 (UTC)

perpendicular recording

Does anybody have a more detailed description of what perpendicular recording is? I checked over at Valdemar Poulsen page and i couldn't find anything about his discovery as this paragraph from the linked article asserts. "Perpendicular recording was pioneered by the late 19th century work of Danish scientist Valdemar Poulsen, who demonstrated magnetic recording with his telegraphone." [1]

Here is a pretty down to earth explanation of Perpendicular recording. [2]

Altitude Sensor?

"Some modern drives include flying height sensors to detect if the pressure is too low, and temperature sensors to alert the system to overheating problems."

I do not believe that as of today, April 2005, there are any hard drives that have an altitude sensor as these are still to costly. Every modern drive has a temperature sensor. Channel parameters are typically adapted to the measured temperature. BruceSchardt

An altitude sensor is not necessary to detect flyheight change—there are other methods of doing so. Of course, the statement above is somewhat misleading. GMW 15:31, 6 February 2006 (UTC)

Start Stop CSS Testing

I changed the discussion of startstop or CSS testing because the statement that 50% of the drives fail at 50,000 cycles is not correct. Typical specification for a desktop harddrive is 50,000 cycles. However this is not a mean time (cycles) till failure specification. This is a specification that all drives are supposed to achieve without failure. In typical testing a population of several hundred drives are contact-start-stop (CSS) tested for >100,000 cycles. No failures caused by the head-disk interface are allowed at less then 50,000 cycles. Testing for more then 50,000 cycles insures that the HDI design has margin. In proof, the Maxtor Diamondmax 10 hardrive specifies Start-Stop cycles at >50,000 min.[3] Notebook hard drives are all load-unload designs that never park the heads on the disk, instead the heads are unloaded onto a ramp. This is done to increase the magnitude of non-operational shock that the drive can withstand. BruceSchardt RESEARCHER:John Raineil Bayani

---

The article currently says

Most manufacturers design the sliders to survive 50,000 contact cycles before the chance of damage on startup rises above 50%. ... the Maxtor DiamondMax series of desktop hard drives are rated to 50,000 start-stop cycles. This means that no failures attributed to the head-disk interface were seen before at least 50,000 start-stop cycles during testing.

This makes it sound like Maxtor uses a different, more stringent test, and that their hard drives (or at least the sliders) are more reliable than "most manufacturers". Is that really true? I wish I had a reference that explained what is really meant by "at least 50 000". Does it mean that after 50 000 cycles, at least half the drives still worked? Or does it mean that after 50 000 cycles, *all* the drives still worked? --DavidCary 05:59, 1 December 2005 (UTC)

A quick Google for hard drive reliability gives me:

• http://www.storagereview.com/guide2000/ref/hdd/perf/qual/specCycles.html mirror: http://www.pcguide.com/ref/hdd/perf/qual/specCycles-c.html
• "Hard Drive Buyer's Guide" http://www.pcbuyersguide.com/hardware/storage/harddrivebuyersguide.html
• "Estimating Drive Reliability in Desktop Computers Consumer Electronics Systems" http://www.digit-life.com/articles/storagereliability/
• Hitachi Focuses On Improved Hard Drive Reliability 2004 http://www.techbuilder.org/news/59201469
• "Digital Dark Ages?" http://slashdot.org/article.pl?sid=02/07/11/058229
• Hard Drive USA http://harddriveusa.com/

Are any of these worth sticking into the ==External links== section? Feel free to delete them if they are not.

Form factor

There is no reference to the different form factor HDs: 3,5 and 2,5

And 5.25" and, of course, various custom, legacy and/or historical forms!

And please add more pics of old hard drives too!

There IS mention of sizes under "Other Characteristics". I added mention of 0.85", for example. 5.25" HDDs haven't been used in a while, so they're not as important as getting current information right, in my opinion. (I have a couple old full-height 5.25"ers stored away someplace....)

This whole article needs to be split into parts and made coherent. Until then, it gets jury-rigged ;-) GMW 07:26, 29 April 2006 (UTC)

I'm thinking of spending time gradually copying information into a new History of HDD wiki page. It'll be duplicate with this page for awhile, at least until some level of apparent common consensus indicates the extraction & moving is approved-of.

Any suggestions or advice would be greatly appreciated. What concerns me most is, does this article provide the information its audience presumably anticipates? The history in particular seems specialized, hence my plan to move it to a history page. I think the "Other characteristics" section should be renamed "Characteristics" and moved up above the mechanical details; isn't it better to overview the properties and then delve than vice versa? Geo.per 20:02, 26 June 2006 (UTC)

I agree that the history of hard disk drives should be split into a new article because a)it is an important subject and b)it is getting pretty long and making the hard disk article too big. My suggestion would be to make some kind of a timeline from the very first hard disk to the newest one, mentioning when certain technologies like IDE and S-ATA begin to be utilized. A capacity graph would be interesting (kilo-mega-giga-tera-etc) --Boneka 19:21, 16 September 2006 (UTC)

Dangers

I don't wish to rain on your parade but hard disks crash no matter what technology has been produced to prevent this.. I would wish that labels be put on iPod's to notify the users of the iPod's that in certain ccircumstances jostling the player around could crash the hard disk and render the player useless.. Maybe someone can tell me differently.. --Rofthorax 08:13, 15 August 2005 (UTC)

Do not degauss your hard drive. --Ihope127 18:29, 16 August 2005 (UTC)

Well, that's not entirely true. HDDs made for portable CE devices typically have a shock sensor or drop sensor that allows the firmware to detect if the device has been jostled or chucked toward the ground at high velocity. The firmware then unloads the heads. (On the other hand, what happens after the drive bounces off the ground is an entirely separate concern!) Furthermore, portable music players usually buffer data in memory and then unload the heads so as to minimize the risk when the people in those iPod "silhouette" commercials do their shimmy. Of course HDIs are caused by other mechanisms as well, such as particulate contamination hitting the head or from deformation of the ABS during start-stop. Hey, no technology's perfect, especially in the volumes HDDs are produced in. GMW 18:05, 4 February 2006 (UTC)

This article needs a section on the dangers of magnets to a hard disk's operation. What the effects are of bringing a hard disk near a magnet (if any). --Umma Kynes 20:21, 12 August 2006 (UTC)

Noisy Chatter

Can anyone here add to the page why EXACTLY do hard drives make the sound that they do when transferring data?

The EXACT reason is because various things inside the drive are moving around. --Brouhaha 08:52, 14 December 2005 (UTC)

The noise comes from seek operations where the VCM moves the head over the track containing the information to be read or written. When one track is read/written, the head must be positioned at the next track, and so on. Depending on the areal density and radius, the number of sectors per track differ, but the range is generally 300-1200, well under a megabyte per track. If one expects the HDD to have a transfer rate of hundreds of megabytes per second, the actuator will be moving frequently, as the head jumps from track to track at the rate of hundreds of operations per second. (Of course interface transfer rates are an upper bound and differ from the media transfer rate, which also varies by the above factors.) Actuator movement on modern HDDs is minimized by switching heads up/down the head stack first, then seeking to the adjacent track. GMW 17:55, 4 February 2006 (UTC)

Minor clarification and addition to my previous comment. Since the movement of the actuator happens hundreds of times per second, the mechanical vibration is modulated in the audible frequency band, i.e. hundreds of Hertz, and can thus be heard.

There are three methods used to handle physical-to-LBA mapping. One is to use one surface entirely, then switch to another surface, but this isn't used on any modern drive that I know of. Another is to go from one side of the disk to the other (usually outer-diameter to inner-diameter), alternating surfaces before moving to the next track. Last is a hybrid between the previous two, moving across one surface in a zone or sub-zone area before switching surfaces. The latter method has many different variations, whose merits depend greatly on the servo characteristics and usage application of the device. GMW 07:57, 29 April 2006 (UTC)

Hard disk != Hard disk drive

Hard disk == the spinning CD-like thingy Hard disk drive == the hardware that reads and writes to the hard disk

just thought I'd point it out :)

Semantics, semantics =) The old IBM Storage Tech term for HDDs is "file". Veteran engineers still use the term, but newer engineers just say "drives". "HDD" is too unwieldy. (I suppose I'm a bit surprised that no one from Seagate has adjusted the spellng of "hard disk" to be "hard disc", which they seem to prefer.)

HDD media isn't really CD-like, since the substrate is aluminum or glass and not polycarbonate plastic. Nor does it exhibit the interference spectrum, as there are no pits and lands; an HDD platter actually looks like a mirror because of its exteme flatness and plated magnetic surface. One of my coworkers keeps a 2.5" disk on his desk propped up by one of those binder clips (black metal spring clip with chrome latches) for preening =)

I do agree, though, that this article should be called "Hard disk drive" rather than "Hard disk", as that's technically more correct. GMW 16:52, 25 March 2006 (UTC)

Perhaps silence can be assumed to be passive agreement? Ardric47 00:41, 16 April 2006 (UTC)

Rewrite

The "Mechanics" part of this article is full of repeats. Somebody should clean it up. --Nappilainen 20:12, 26 January 2006 (UTC)

The "Mechanics" part should contain some diagrams showing the conception of hard disk similar to these in Cylinder-head-sector section or http://www.rwc.uc.edu/koehler/comath/42.html because now it's not very clear what each part does from looking into the pictures.

2005 - 1 TB hard drives

2005 - 1 TB hard drives

Any references to this? Last Avenue 00:52, 27 January 2006 (UTC)

• I can see how this happened. Some companies are providing an option of 1TB as a primary Hard disk. They are in fact 2 500GB drives in RAID 71.213.27.179 16:55, 11 September 2006 (UTC)

Derivative Technologies

I removed the sentence and the section regarding derivative technologies. The sentence was poorly written and confusing. That could be fixed, though. The real sticking point is the lack of references or citations to support this allegation. If it's true that floppy disks and CD-ROMs are derivatives of hard disk technology then this fact certainly needs to be mentioned - with supporting references. --ElKevbo 14:39, 2 February 2006 (UTC)

Other characteristics

I edited Physical Size and it became a huge blob of text. Perhaps I should've bulletized each form factor.

It seems odd under MTBF to put that SATA supports 10k RPM and FC supports 15k RPM. These things aren't directly related. While spindle speed does affect BPI and thus the data rate selected for a given zone, one could just as well have a 10k RPM PATA drive. MTBF itself is a very difficult topic.

Perhaps it would be useful to glean all the parameters generally published in datasheets, so each one can be explained for the reader.