Harddisk Density vs Data Retention Time ???

This _can_ be an issue, for several reasons, and it's well-known in the hard drive industry. I had some discussions with engineers at a couple of major hard-drive manufacturing companies about this, about ten years ago when I was working in the DVR business.

The smaller magnetic domains do have poorer data retention - they're more subject to being affected by thermal fluctuations which "flip" the magnetic orientation of individual atoms. Over time this can result in the data becoming harder and harder to read.

A bigger problem (one which happens sooner) arises from "fringing fields". Although read/write heads have gotten smaller over the years, it's not possible to keep the magnetic field from the write/erase heads tightly enough "aimed" to affect only the data track you're writing to. Some of the field's fringes reach out far enough to impinge on the adjacent tracks, so (for example) every time the head writes new data to track 100, the write fields have a slight effect on the magnetic domains for tracks 99 and 101. This is a cumulative effect, and over time it can weaken or erase the data on the adjacent tracks.

The hard drive controllers for drives like this are designed to manage the problem. They keep a record of the number of times each track has been written to. Periodically they'll check the record, figure out which tracks have been written to more than a certain number of times, and then go check the data quality on the adjacent tracks. If a track's low-level bit error rate has gotten too high, the drive will copy the data to a spare area, then rewrite it back into place.

I was told that when planning out the "use time" budget for a drive of this sort, when used in an "almost always reading and writing" application like a TV DVR, I ought to reserve 10-15% of the total time for this sort of internal drive maintenance activity - health monitoring, and data refreshing. There's no additional computer or DVR work required for this - the drive controller does it "automagically" - but you have to plan to leave enough "idle time" for the drive to do the maintenance in the background.

An extension of this issue involves the new "shingle-mode" recording, in which the physical tracks are so close together that they partially overlap. Drives which have shingle zones don't rewrite data in place... these zones are treated more like solid-state drives, where the drive always writes new data into a newly-erased zone, and maintains a virtual-to-physical mapping table for each sector. That way, a shingle sector will only be affected by partial-erasure fields once or twice, when the next-inbound sectors are written.

snipped-for-privacy@coop.radagast.org (Dave Platt) wrote in news: snipped-for-privacy@coop.radagast.org:

SkyTard Trolling is an idiot.

And you must not be keeping up either show me a link to your "new shingle mode' crap.

I have one for you. It is called perpendicular recording and is the reason why you can have terabyte sized hard disks which are only

2.5 or 3.5 inches form factor or smaller. Been around for a long time. Even before the hard disk tech leader IBM sold their HD segment to Hitachi they came up with perpendicular tech. Those domains are VERY CLOSE together.

Every disk manufacturer is doing it for large drives; check out WD site for 'SMR' shingle mode recording disks, like every single RED drive...

But what about erasure? You cannot generally erase/rewrite any single bit without clobbering others nearby, so it is vital to have a format for the data that allows blank space between allocation blocks on a single track, and between tracks on a disk unless one 'shingles' the tracks and erases/rewrites multiple tracks, taking several disk revolutions to complete every single (long, many-bytes) write.

It's OK to put some random garbage into an utilized gap. WIth shingled access, the gaps are still there, just MUCH farther apart. The minimum-data-write blocks are accordingly very large.

[snip]

I've noticed for some time that my HDD activity led comes on when nothing else is going on. I assumed it was a defrag or some such operation.

you mean large drives where terrible performance doesn't matter. shingled drives don't cut it for operations where writes have to happen in a reasonable period of time.

Cydrome Leader wrote in news:s19qs1$q3f$ snipped-for-privacy@reader1.panix.com:

At over 40Mbits per lineal inch on a platter drive, they get written to just fine.

They are at up to 5Tbits per square inch.

Only idiots buy slow 5400RPM disk drives.

You're not qualified to refers to others as "idiots" when it comes to current hard disk drives, performance and technology.

I'd guess that is was... or a virus/malware scan, or an in-the-background software update, or something else being done by the host operating system.

In most systems, the HDD light is driven by the motherboard (e.g. the hard-drive interface, typically SATA these days). It only comes on when the OS or BIOS accesses the drives.

Operations which are internal to the drive itself (generated by the on-board controller) won't light this LED. If the drive itself has an LED (few seem to, these days), the LED might or might not come on.

If you _hear_ the drive operating, and the main LED isn't on, there's a good chance it's doing some sort of internal maintenance operation or health scan, all on its own.

Different tasks have different requirements, and the optimal solutions are different.

Shingled drives are optimized for certain sorts of applications. They could work well in some digital video recorder applications, where writes are fairly large, often sequential, and where you can do big reads and buffer up a bunch of video before you start to play it back. Similarly, they're suitable for some home NAS applications - music and video libraries, for example. They can be a good choice for backup applications, depending on how your backup system operates. Due to their high track density, I imagine that they can deliver a larger amount of storage per dollar than other technologies.

As you correctly point out, they're poor performers in applications which need fast and predictable random access... databases and other object stores, software-development workstations, many sorts of Web server. For these, more traditional hard drives (or SSDs) are a better choice... less storage per dollar, but more random accesses per dollar.

Is a $150,000 tractor-trailer a better motor vehicle than a $150,000 sports car? It depends very much on what you want to do with it.

Cydrome Leader wrote in news:s1bhg0$drf$ snipped-for-privacy@reader1.panix.com:

I definitely am. I use hybrid drives and have never used optical discs for archiving.

You are not qualified, nor were you solicited for your umprofessional opinion on what I may or may not be qualified to do. Your domains are convoluted, boy. You lead nothing.

You are:

UNqualified UNsolicited UNprofessional and UNeducated on anything I know.

You mumble little dribbles out of your mouth just like Trump does.

Sorry you don't have any real experience. Keep trying though.

I take you don't actually work with some of the highest performance storage systems on the market. So, just got ahead and play with your trash from the best buy returns dumpster.

Sounds like you got real upset there, real fast. Way faster than a POS high latency shingled drive can erase and rewrite all its tracks over 1 byte change.

They are only optimized for cheap cost for storage capacity. There is nothing optimized about them when it comes to realiability or performance.

dollar for dollar shingled drives always have absolutely terrible performance compared to conventional drives. The don't offer more torque or haul more stuff like a tractor vs. a sports car.

[about shingled hard disk drives]

That's not completely true; the write speed is identical to conventional drives until you exceed the RAM buffer, and the read speed, because there aren't as many gaps, will be higher for large files. As an archive, or backup drive, they don't need any better reason to exist than cost per bit.

I love the qualifier of "until you exceed". shingled drives are too busy tripping overthemselves to offer even slightly acceptable random write speeds. That's not even getting into command queue depts or timeouts that can occurn on a read that's held up from previous writes. There's plenty of stories about how these drives perform so poorly they can't even be used in an array. There was also recent "scandal" about how WD forgot to tell people a line of drives were shingled. They finally fessed up as there was no way to hide how terrible the things were. Here is the story

Maybe they're passable if you need a cobbling together a VTL or single camera video recoder and have time to untune every setting in your storage controllers, or just don't care if you lose the data.

OK, I will assume for a moment what you wrote is true. Though I don't know if hitachi harddisk do this ?

But let's assume they do, or harddisk like this, whatever.

Here is another question about this:

Would the harddisk led become bright when the drive does this ? Or does it do it more sneaky ?

Bye for now, Skybuck.

Reading rest of thread now, apperently already answered, apperently it doesn't at least not on case, which replier claims is driven by motherboard/bios... Hmmm... this is a little bit concerning ! ;)

I also wonder if windows takes this into account somewhat, probably not, but usually defrag only start after 15 minutes or something.

Bye for now, Skybuck.

e:

er

l

-recording/

e

Wow I might have actually bought one of these at the media markt.

It's right now in my PC, I did install a fresh version of windows 7 on it. I didn't really notice any performance problems or so.

I am not even sure if it's a shingled drive, but it is a WD RED 3.5" inch d rive.

The packaging claims it has optimized RAID performance.

I wasn't even sure if this NAS drive was going to work in my PC, it's 2 TB to work with the 2006 bios from winfast motherboard.

The packaging is red and white like on that link but looks a bit different.

I found a model sticker on the bottom of the package which I thankfully and ofcourse smartly kept, which I always do in case I have to return broken o r defective stuff ! LOL. The sticker has very tiny letters on it. It smells kinda fishy that this was not printed on the packaging itself. Almost like WD wants this sticker to go loose and missing ! ;)

Model: WDBMMA0020HNC-ERSN PRODUCT OF THAILAND

3 YEAR LIMITED WARRENTY.

I have some digging to.

Bye for now, Skybuck.

Here is a statement from WD about it:

It still doesn't tell/indicate how to know if the drive I bought has this t echnology.

I hope there is some information on their website which I will try and visi t next.

So far this 2 TB drive I have barely used it, just installed windows 7 as a backup OS and as a gaming os for Geforce now.

Hmmm...

Perhaps the RED packaging is a BIG FAT WARNING that something is wrong with these drives ! LOL.

I always listen to my GUT FEELING and I kinda had a doubtfull feeling about this drive, something was telling me there is something off or perhaps dan gerous about these drives.

It took me months before I installed it ?! ;) and I am reluctant to use it ! ;) =D

Also strangely enough my older windows version does not detect the label of the drive correctly, though that might be a little bit strange windows bug in an older version of windows 7 ?

I think I performance tested it and noticed nothing bad about it or perhaps I am delirious and didn't test it yet hmmm... nope I tested my usb portab le seagate drives benchmarked those... well time to benchmark this RED driv e to see if I can find something odd about it ! ;) :) Hopefully it don't ca use problems ! ;) =D But rather now than later ! LOL.

Bye, Skybuck.

I benchmarked/performance tested my Western Digital NAS Hard Drive for the first time with my older windows 7 installation, with ATTO storage/harddisk benchmark tool. Same OS and settings as I tested all my other harddisks wi th.

So far this model I wrote about is the fastest harddisk I own, it completel y blows the other harddisks out of the water for smaller little blocks, for example 512 bytes or 1 kilobyte blocks.

Why this drive performs so well for tiny little blocks I don't know, maybe because it's unfragmented/little used, or maybe it has a very big read cach e / RAM chip for caching or perhaps it's optimized for small little reads. I have no idea.

But it can do 11 Megabyte/second read and 13 Megabytes/sec for 512 blocks, which is quite impressive. The other harddisks all struggle with such tiny blocks and can barely do 1 Megabyte/sec or 3 megabyte/sec at best.

I think I felt this snappyness a little bit when I installed the new window s 7 version.

The top write speed and top read speed falls a little bit lower 112 megabyt es/sec all across the board from 8k to 16kb blocks and up, very consistent read and write speeds. So far no performance issues detected, on the contra ry, this is basically the fastest harddisk I own ! LOL.

I could try benchmarking old harddisk from pentium 166 and 80486 from long long long long ago, but I doubt they will be any faster than this 2 TB driv e ! ;) But who knows, maybe old harddisks where faster in those days, I rea lly dont know =D

For now I am quite pleased to own this one, though I don't use it much yet, maybe in the future ! ;)

Still gonna look into it to see if it's a shingled drive, I have no idea ri ght now.

Bye for now, Skybuck.