Jump to content

Talk:Shingled magnetic recording

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Overlapping tracks

[edit]

This sounds similar to the technology, that LS-240 (Superdisk) drives used to write 32MB on 1.44MB floppies with the standard 1.44MB floppy head (FD32MB). Should this similarity be mentioned in the article? FD32MB is explained here in Japanese, which I don't understand, but the graphics show in principal how it works (the "125 µm" must refer to the flopy head, which is used for writing and the "8 µm" to the Superdisk head, which is used for reading). --MrBurns (talk) 01:47, 24 February 2015 (UTC)[reply]

I'd prefer if there was a source saying that the techniques are similar, otherwise it's WP:Synthesis. -- intgr [talk] 02:04, 24 February 2015 (UTC)[reply]
Hello! Based on an English translation of the explanation and the image included there, it seems what's desribed is the difference between "floppy" and "LS-240" heads and resulting track widths, not the overlapping nature of recorded tracks. Actually, if I'm not wrong the image would suggest that tracks overlap on ordinary 80-track floppies, what certainly isn't the case. Here's an excerpt from the linked translation:
Improvement in track density, in writing at the time of conventional 2HD head (125 μm), and by recording while shifted by 18.8 μm, the track width 18.8 μm, formats the tracks 777 this one side. Read at the time I use the super head for disk of 8 μm width. As a result, the number of tracks is increased to 777 tracks from one side 80 tracks, storage capacity is increased about 9 times.
If the tracks depicted on the left side in the image were the 18.8 μm tracks made by "LS-240" recording heads, that would indicate their ovelapping nature, but they're clearly marked as 125 μm tracks so it isn't the case. On the right part of the image, it is clearly depicted that recorded 18.8 μm tracks are read with 8 μm "LS-240" read heads (the "I" is narrower than the track). The only remaining confusing part is that 125 μm on the left is completely disproportional to 18.8 μm on the right. — Dsimic (talk | contribs) 02:29, 24 February 2015 (UTC)[reply]
Actually if you install the Super Writer 32 utility (which is used at least for Panasonic/Matshita LS-240 drives) on Windows, you get an English .hlp file (thee files can still be read in Windows Vista/7 if you install a viewer from Microsoft), which contains a section "Features" section. In this section it is explained in detail how it works. According to this the tracks overlap at writing (which is done with an ordinary HD-floppyhead, however because each time a new track is written a part of the old track is overwritten, so after writing there is no overlap left. For reading the LS-240 head is used. This explains the difference between 125µm and 18.8 µm. The Super Writer 32 utility can be downloded here: [1] --MrBurns (talk) 23:47, 24 February 2015 (UTC)[reply]
Any chances, please, for a few screenshots? Moreover, those .hlp files should actually be plain HTML packed in some fancy format, IIRC, so it should also be possible to convert them into HTML pages, host them somewhere, and use as references.
Also, I've tried to find more online references, and these two seem to be usable:
However, none of them mentions that the 125 μm "floppy" head is used for writing tracks. — Dsimic (talk | contribs) 00:25, 25 February 2015 (UTC)[reply]
I think about uploading a screenshot of the relevant information (should fit on one screen on my PC), but I am not sure if this is legal (because of copyright). The floppy head is only used for the FD32MB format, which writes up to 32MB onto a normal 2HD floppy disk and is only supported by LS-240 drives, not LS120 (and of course the floppy head is also used for normal floppies). For Superdisk, of course only the Superdisk head is used. --MrBurns (talk) 08:23, 25 February 2015 (UTC)[reply]
You can use any free image hosting service (imgur.com, for example) for the purpose of exchanging information on talk pages. We'll see what to do later, if we decide to use that information as a reference. — Dsimic (talk | contribs) 08:29, 25 February 2015 (UTC)[reply]
Ok, this screenshot contains the relevant "Improve track density" part of the "Feature" section. --MrBurns (talk) 08:46, 25 February 2015 (UTC)[reply]
Thank you for the screenshot! In the beginning, it clearly says the following:
The magnetic head of SuperDisk drive consists of two parts. One is "Writing and Reading head" (125 μm width) for floppy disk. Another head is "Writing and Reading head" (8 μm width) for super disk.
That would indicate that the 8 μm head is used for both read and write operations on SuperDisks... But, that doesn't seem to be true, as it contradicts with the illustration in point three (it's the same as the image in the earlier source in Japanese, but now we can read the labels :), which indicates that the 125 μm head is used with SuperDisks for writing data, while the 8 μm head is used for reading data. Also, the point two says that the 125 μm head overwrites tracks to create narrower "leftovers" that are later read using the 8 μm head.
With all this, I'm pretty much convinced that the SuperDisks used shingled-style recording. What do you say, intgr? — Dsimic (talk | contribs) 09:11, 25 February 2015 (UTC)[reply]
Actually the Superdisk drives don't use shingled recording when writing on superdisks, only when writing on standard floppies. The helpfile from which I took the screenshot is for the FD32MB format, which is used to write 32MB on a standard floppy. Superdisks are written with 120MB or 240MB. --MrBurns (talk) 09:17, 25 February 2015 (UTC)[reply]
PS: a LS-240 superdisk drive can write the standard floppy formats on standard floppies, it can write the FD32MB format on standard floppies, or the LS-120 or LS-240 format on superdisks. If it writes or reads standard format floppies, it uses them like a normal floppy drive (i.E. only with the floppy head), for FD32MB, it uses the floppy head for writing and the superdisk head for reading, and for LS-120/LS-240 it only uses the superdisk head. --MrBurns (talk) 09:22, 25 February 2015 (UTC)[reply]
Ah, sorry, I got a bit confused, thanks for the clarification. This source just confirms the SuperDisk's ability to format and use ordinary floppies to achieve 32 MB of capacity. — Dsimic (talk | contribs) 09:34, 25 February 2015 (UTC)[reply]
Is it not possible to just use the SuperWriter32 helpfile as source? It is part of the setup of SuperWriter32 which can be downloaded here. The setup installs correctly even on computes without Superdisk drives, only in this case the SuperWriter32 application will stop with an error message if you try to run it, but you can still open the .hlp file, which is lokated in the program directory. The helpfile is called Spwt32.hlp, but I cannot find it with google, so downloading and installing the whole setup is the only way I know. The program is from Matsushita (Panasonic), one of the biggest LS-240 manufacturers. --MrBurns (talk) 19:20, 25 February 2015 (UTC)[reply]
I'd suggest that you convert that .hlp file into .chm format (using this utility, for example), then convert the .chm file into plain HTML (using this utility, for example), and host the resulting HTML somewhere so it can be used as an online reference. I'm not 100% sure whether it would even be acceptable to create a PDF file out of the resulting .chm or HTML, and upload it to the Wikimedia Commons? Could you, please, check that out? As a note, there might be much better utilities for that purpose, these two are just the result of using Google for about 20 seconds. :) Speaking of references in general, we even try to avoid websites that require login, so requiring someone to install an application wouldn't be that great. — Dsimic (talk | contribs) 23:16, 25 February 2015 (UTC)[reply]

@Dsimic and MrBurns: Sorry I didn't reply earlier. I had a look at the sources and the technologies certainly seem similar to a storage doofus like me. But it requires a fair bit of interpretation and there may be more to SMR than seems on the surface. Anyway, I think it would be a nice thing to add to this article, just don't claim that the two technologies are same thing since we don't have a good source for that. Stating "similar" should be fine. -- intgr [talk] 12:45, 3 March 2015 (UTC)[reply]

Thank you, intgr, and no worries about the delay, there's no big hurry. Sure thing, we should describe it as a similar magnetic recording technology dating back from the early 2000s – it's by no means the same thing as SMR. MrBurns, have you managed to check whether uploading the above described PDF file to Wikimedia Commons (or locally, to English Wikipedia) would be Ok with the licensing requirements? — Dsimic (talk | contribs) 04:17, 4 March 2015 (UTC)[reply]

Deleted SMR Glossary of Terms

[edit]

@Intgr: I would really like to undelete the list of terms that user:intgr deleted: [2]. While I am no expert on shingled magnetic recording, the list contains a lot of what seems like really important definitions, and Wikipedia articles should include definitions. user:intgr alluded that the list should be rewritten as prose, but to me it seems that it is perfectly adequate in list form. And at the very least, if editors still think prose is better than a list, the section containing that much useful information should be tagged for rewriting instead of outright deleted. Deleting that much useful information outright does not make Wikipedia better, IMO! Thue (talk) 19:00, 7 September 2015 (UTC)[reply]

Hello! Well, glossary sections really aren't what Wikipedia articles should contain. However, some of the information available in the deleted list of terms was really useful, and I'd suggest that we restore one part of the list as prose. Please keep in mind that we don't need all information to be restored in prose because one part of it is already explained in the article. Also, we'd need additional references. — Dsimic (talk | contribs) 20:01, 7 September 2015 (UTC)[reply]
I don't necessarily see why an article on shingled magnetic recording shouldn't contain a glossary of terms relevant to shingled magnetic recording, if it is informative. Can you quote a policy article on that? I know it is not very common, but I don't see the problem with it. Wikipedia:Wikipedia is not a dictionary only talks about not making the whole article into one dictionary definition. Is it any different than an article with a list of minor characters in a TV series? - Name/explanation list-form info such as that is not unusual in Wikipedia. Thue (talk) 16:22, 8 September 2015 (UTC)[reply]
Just revisited our (numerous) guidelines, and the MOS:GLOSSARIES guideline covers glossary-style sections and articles. It clearly says that having glossary-style sections in regular articles is fine, but I'm not sure whether the guideline itself is accepted as part of the Manual of Style? Thus, we seem to be clear to have a glossary as part of this article, but I still find that converting it into prose would actually be much more useful; such a long glossary-style section would dwarf the article and turn it visually into something that resembles a glossary-style article with a brief lede, which the article surely isn't to be. I'm not sure if I'd have time to do that, but I might take a shot at it; after that, we could discuss further whether having a separate glossary-style section would still be needed. Thoughts? — Dsimic (talk | contribs) 23:03, 8 September 2015 (UTC)[reply]

Needs clarification

[edit]

I'm not sure what this sentence is saying: "The overlapping-tracks architecture may slow down the writing process since writing to one track overwrites adjacent tracks, and requires them to be rewritten as well."

This sounds like a cascade of writes! Write one track, overlapping track 2, which then needs to be rewritten, overlapping track 3, and so on... — Preceding unsigned comment added by Jlowery2663 (talkcontribs) 02:15, 28 February 2020 (UTC)[reply]

It is indeed such a problem -- writing track n corrupts track n+1, which cannot be fixed by rewriting track n+1 because that corrupts track n+2, etc. Device-managed SMR must avoid this by turning the whole device into what is effectively a running log of changes to any data, always adding any changes to the end of the last track written, instead of simply overwriting old data with new in situ like normal drives; the device must then constantly weed-out the obsolete (old) data in the background to reclaim space for more logging. Even if this is well-managed by the device to do this only during lulls between write commands, it still costs some performance on average. On host-managed SMR, the host knows what data it is corrupting with new writes, so it only selects tracks with what it knows has only obsolete data for corruption, minimizing but not eliminating the problem.
Thus, yes, the sentence is correct, but per the post, this isn't stated clearly enough.
So, how do we make this more clear? --A D Monroe III(talk) 02:12, 29 February 2020 (UTC)[reply]
I did not find the original particualrly problematic but I've taken a shot at improving it ~Kvng (talk) 15:01, 2 March 2020 (UTC)[reply]

Divice managed section

[edit]

The third paragraph is a mess. Lots of assertions without citations. Why is the rebuilding of a RAID mirror problematic, when it should be an entirely sequential write? 87.75.117.183 (talk) 10:57, 25 February 2021 (UTC)[reply]

PMR and CMR

[edit]

PMR and CMR are not the same thing. Perpendicular recording describes the way that bits are recorded whereas Conventional recording describes the way that tracks are recorded. Therefore, an SMR drive is also a PMR drive, ie the terms are not mutually exclusive. In a typical SMR drive, the bits are recorded vertically (PMR) while the tracks are shingled (SMR).

Which Drive Has What?

https://www.seagate.com/au/en/products/cmr-smr-list/

Determine if an Internal Drive uses CMR or SMR Technology:

https://support-en.wd.com/app/answers/detailweb/a_id/50697/~/determine-if-an-internal-drive-uses-cmr-or-smr-technology

106.69.100.203 (talk) 05:13, 9 August 2023 (UTC)[reply]