Western Digital's 16TB and 18TB drives: EAMR HDDs enter the retail channel

https://www.anandtech.com/show/15903/western-digitals-16tb-and-18tb-gold-drives-eamr-hdds-enter-the-retail-channel

by rbanffy • 6 years ago

117 164 6 years ago

anandtech.com

smart_jackal• 6 years ago

Whoever has the need to store that much data?

vmception• 6 years ago

2002 called and wants its argument back

oliverobscure2• 6 years ago

Archivists and people who enjoy large collections of data, who are providing a useful service to preserve internet history.

Hamuko2• 6 years ago

Can't think of a single company that might want to store 18 TB on a single disk? These are enterprise-class products after all.

memco• 6 years ago

Even for home use, if you have a family who wants to keep local backups of multiple devices. A phone, laptop and tablet per person for even two people would eat through a few TBs every few years.

peterangular2• 6 years ago

Working on a security solution right now and I can 100% tell you I can think of multiple companies that need an on-site 18TB datastore, even "small" companies. Also have worked with tons of creative agencies. Video, photo, etc all can take huge amounts of disk to safely store.

Sorry to disagree but your assumption that these are "enterprise-class" products is incorrect. There are lots of business needs that require large on-prem storage solutions without being an "enterprise-type" business.

Edit: downvotes? Really? I'd really love to see someone make a case against this.

Spooky232• 6 years ago

It's definitely a thing. Media is one use case, logging is another.

pheug2• 6 years ago

Very useful for storing medium/large-scale crawls. Common Crawl is 200TB+ per month (http://commoncrawl.org/connect/blog/). Uncompressed, if compressed it'd probably go through one such disk every 1-2 months. My own crawling operations consume about half a TB each month.

Storing high resolution (1080p/4K) videos is another big spacehog

selykg• 6 years ago

I have something like 5TB of tv shows and movies on a Plex server since I hate swapping in DVDs and Blurays for the stuff I watch more than once and my collection isn't even really that big.

But anyone with a large video library could easily use up a ton of storage. I've filled up my external drive used for Plex, it's upgrade time for me and I'm going for a 4x8TB NAS and hopefully that'll keep me running for at least a few more years.

peterangular2• 6 years ago

Software engineer who deals with large datasets, disk images, etc etc. Throughout the years I've had to have a big "scratch disk" with yet another bigger "archive" to support my work.

For my scratch disks I need at least 1TB locally (preferred 4+), and for my archive I need 16TB easy. I deal with everything from shape files, to GIS, to huge flat-file processing etc etc. I'm the guy who can take 10+ years of data and make business sense of it so storage is a constant headache/need.

To get ahead of someone telling me, "you're doing it wrong": no. No I am not. Just because you don't have the need doesn't mean there's not a market for it.

NikolaNovak2• 6 years ago

It's really not that hard.

If you're a photographer for example, storage gets filled up VERY VERY quickly. A wedding photographer may do 1000-6000 photos in a weekend, 40-100MB each; which is 40GB - 500GB per event, depending on your workflow.

Video even more so - whether it's an entertainment library, or high quality videos of family, let alone if you do it professionally or semi-pro.

Home labs for anything - virtualization, data science, ML, CI/CD, etc.

My current NAS has 8TB drives in it, but only because of the price curve. I would've preferred 12TB but 8TB is where the price/capacity/performance cusp seems right now.

AuryGlenz• 6 years ago

Yep, I’m a photographer and I’ve got 4x8TB drives in a NAS and it takes some work to make sure it doesn’t fill up. I like to keep all my wedding raws for a year, and I keep all the raw files I haven’t culled long term.

commandlinefan• 6 years ago

Anybody who wants to run the latest CoD render pack.

doublerabbit• 6 years ago

Animation studios. I worked for one that produced content for Netflix and they were struggling with space issues. With just over 1PB of storage data, it gets tricky to fit more hardware.

You have to keep archival data in case the studio decides to comes back, not forgetting you have to keep revisions, the raw formats, the rendered formats, the render raws, audio raw, audio mixed.

Television Studio's and Broadcasting Companies all occupy lots storage too.

dhosek• 6 years ago

When I bought my first computer it had a 200MB hard disk drive. One of my former professors told me I was nuts, that I would never fill that much disk space.

iso8859-1• 6 years ago

I remember people complaining about the quality of WD drives. These ones claim a MTBF of 2.5M hours, workload rating of 550TB/yr, and a 5 year warranty. Can it be trusted?

murgindrag5• 6 years ago

No. It absolutely cannot be trusted.

I buy drives in pairs, and RAID them. If you can't afford to do that with drives the size you need, buy smaller drives and keep less data. It's like being a half-decade behind on the Moore's Law curve. You get used to it and survive.

greenshackle2• 6 years ago

Do you buy pairs of identical drives? If you want to be extra paranoid, you can use a pair of different models or at least different batches to reduce the risk that they both fail in a short time window.

mceachen7• 6 years ago

RAID won't save you from bit rot, and after I added bit rot detection in PhotoStructure, I was surprised to see how many of my files on older spindles had succumbed to minor corruption.

Using a filesystem that supports data scrubbing doesn't require an "enterprise class" budget anymore. After fielding this question from several of my beta users, I wrote it up the why's and how's on my blog: https://photostructure.com/faq/how-do-i-safely-store-files/

deelowe• 6 years ago

With larger drive sizes and failures tending to follow a rather normal distribution w.r.t. mtbf, raid is becoming less of a safeguard these days. There are some good calculators online that will do the math for you, but at these sizes, you're very likely to see the second and even third redundant drive fail before the rebuild is complete.

SamuelAdams2• 6 years ago

Well if the warranty is any good you don't have to trust them. If the drive fails under expected conditions within a 5 year window, execute the warranty policy and get a replacement.

dannyw4• 6 years ago

Anecdotally, as well as supported by Backblaze, Seagate drives fail more often than WD.

https://www.backblaze.com/blog/backblaze-hard-drive-stats-q1...

I’ve had one Seagate drive fail, and another that would actually return corrupt data even immediately after being written. Thankfully I have ZFS which is how I noticed it in the first place.

I’ve owned more WD drives for longer and I’ve actually never had one fail; at least before they store so little I graveyard them.

lukevp2• 6 years ago

Same experience here, only seagate drive failures over the past 15 years of owning maybe 30 drives total. WD has never failed. Obviously anecdotal.

Mindless2112• 6 years ago

My own anecdote is about the same, except in my ZFS-based NAS, I've had one WD drive fail. It was a WD Green though, so no surprise.

FabHK• 6 years ago

The way I read the Backblaze stats is opposite - could you clarify?

* most reliable are HGST drives (model number starts with H, formerly Hitachi, such as DeskStar, UltraStar, etc., now under the WD brand), sub 1% AFR;

* then Seagate (model number starts with S), still pretty reliable at around 1% AFR, with a few exceptions;

* while the original WD (model number starts with W, such as WD RED etc.) drives are quite terrible with AFR from 2% to 5%.

(I'm looking mostly on the long term statistics, spanning at least a year, and older stats regarding WD drives, as they don't use them anymore.)

rbanffy5• 6 years ago

and a 5 year warranty.

I would bet a 5 year warranty means they trust their hardware. Still, when you have so much data and you really care about it, you should use a filesystem that prevents bit-rot and can deal with multiple drive failures.

And backups. Never forget the backups.

dsr_• 6 years ago

A 5 year warranty means:

the expected failure rate over 5 years, times

the percentage of customers who claim warranty on those drives, times

the future price of an equivalent drive

is less than their general profit margin on the drive.

Let's say that the price is $600, the marginal cost of a drive is $400, and the expected 5 year failure rate is a whopping 20%, with 50% of buyers claiming warranty.

That allocates $40 of costs over the next five years to warranted replacement, against a current profit of $200.

SamuelAdams2• 6 years ago

Would these be overkill for a personal NAS at home?

murgindrag5• 6 years ago

Here's the way I figure:

1) If you're a student, yes. Store less stuff. 1080 instead of 4k video won't kill you.

2) If you're a software engineer, factor in at least $200 of your time for each upgrade. Bigger drives mean you need to upgrade less often, so no, buying more is not overkill.

ubercow132• 6 years ago

Do software engineers really put dollar values on all of their time?

dannyw4• 6 years ago

Yes. Give it 3-6 months and you’ll see perfectly fine but non-spec-rated HDDs sold as EasyStores, etc sold at 40% off.

It will be on the market soon. They will have to sell those binned parts somehow.

rs23296008n13• 6 years ago

Do you have 12TB of data right now? Yes? Then a minimal RAID would need at least two 16TB drives.

dragontamer4• 6 years ago

The important bit for NAS is redundancy. If you can afford 2+ of these drives to run in RAID1, RAID5 or RAID10, then go on ahead.

For me personally, my 5TB x 2 RAID1 (ZFS really, so mirrored) configuration with 5TB usable is enough for my personal uses. If I need more space, I'll configure my ZFS with +2 HDDs for another pair worth of space+redundancy.

duncanawoods2• 6 years ago

If you can only afford 2 then I decided against RAID/ZFS. The issue is that "raid is not backup" and I fear user error as much as hardware error so instant mirroring would be just as bad.

For my home bulk storage, I just have two normal 14TB drives and a nightly cron to rsync archive (new files only) from the primary to the backup. This gives me enough of a backup for both user error and hardware failure. It doesn't protect against a house-fire but that is ok.

I should maybe have considered BackBlaze but I can't believe they are going to last long if they continue offer flat-rates for multi-TB attached NAS.

Latty• 6 years ago

To add to that, one will very quickly find that RAID5 really isn't a good idea for big drives if one reads around.

The reason being that for larger drives, resilvering can take so long that secondary failures become probable or—in bad cases—even virtually guaranteed (This is the same probably the WD SMR drives had, but it exists even on normal drives if they get big enough).

For this reason, for modern large drives, most people recommend at least RAID6 (known as RAID-Z2 on ZFS), which means you have two drives worth of parity as compared to RAID5's single one.

Of course, this means that you need at least four drives for it to be worthwhile, and most people realistically end up going for around six disks, otherwise you are spending a lot of your total cost on parity.

rs23296008n13• 6 years ago

More is better. Those RAIDs are never satisfied.

Its a pity we don't have anything remotely similar for optical storage. Physical limits and all that.

pheug2• 6 years ago

Yeah, what happened to all those 3d optical storage startups that promised us petabyte sized things by now?

mhh__• 6 years ago

Working for "Government agencies"? (Assuming they haven't gone broke)

rs23296008n13• 6 years ago

Possibilities :

Worked & couldn't gain traction. Capital requirements too great. Eg just a lack of funding. Limited prototypes produced but not mature enough. Tech is quietly smouldering while it is passed from company to company. Slow progress.

Worked & got absorbed. Now in some black ops lab quietly still in development with protoypes already in limited use. A limited commercial version is 5, 10 or 20 years away. The restricted version in use within 2-5 years.

Worked & still in commercial development. Fundamental issues are blocking major progress. Possible confusion over usage priorities is diluting resources, eg they are optimising for one or two use cases and those aren't particularly helping guide development, yet those use cases are also where funding is available. 10-30 years.

Don't work & too unreliable. R&D couldnt get a stable and reliable product due to fundamental issues.

Don't work & unreliable & not enough cash. The prototypes sort of worked but had issues and on top there wasn't enough capital to iron out the issues in the early prototypes.

Other options also exist. The timing started 10 years ago. Should start to see results, if any, from the "worked" options sometime this or next decade if that happened and to whatever extent. A lazy web search still shows some names. A few have quietly disappeared.

fnord123• 6 years ago

More is not better. Repair time gets slower as the ratio for transport speed to capacity gets lower and lower. This means that more file pieces can be lost during the longer repair window resulting in possible data loss.

Keep in mind that when people calculate lifetimes and repairs they usually assume that drive failures are independent. Given that they are installed at the same time and usually from the same manufacturer and bought at the same time (e.g. same batch), it's not great to assume that all these things are independent.

manquer2• 6 years ago

You can always buy lower capacity drives and plug more in.

jnwatson• 6 years ago

In my 50 TB RAID 5, it has never finished a monthly scheduled raid sync because it takes roughly a week to complete, and performance during the sync is so bad that it is unusable.

LAMike• 6 years ago

I wonder how many of these hard drives are needed to completely scan the internet every month and create a "time capsule" drive that you can revisit and browse ten years later like a yearbook

soulofmischief2• 6 years ago

The proposed project entails constructing a global-scale shared supercomputer capable of hosting the entire Internet on one platform as an application

https://en.wikipedia.org/wiki/IBM_Kittyhawk

eloff• 6 years ago

So, basically where AWS is headed.

I joke, but there's a grain truth there.

bserge3• 6 years ago

Wow... is IBM the Kodak of computing/hosting?

oliverobscure2• 6 years ago

I suppose one could use the current size of the Internet Archive as a yardstick for how much public-facing stuff there is. Of course in general terms this also depends on your definition of 'the internet' - consider databases and other such stores. Huge numbers.

cbhl2• 6 years ago

If you actually do this, just make sure your crawler doesn't get caught on some programmatic infinite webpage generator.

Like one of those calendar widgets.

manquer2• 6 years ago

With video and other rich media ? You will need hundreds of thousands or even millions of these disks.

YT alone adds 500hrs of video per min . That is at minimum 9000 TB / day (500 18TB disks/day - 180,000 disks a year) [1]

If YT is no more than say 1% of the public internet (Facebook is probably storing lot more , most of the data is not public though, Instagram is ) Then you are looking probably at 18 million disks a year without any backups and copies considered .

[1] it is assuming SD quality 220 mb per hour single format and one copy.

exikyut3• 6 years ago

YT alone adds 500hrs of video per min . That is at minimum 9000 TB / day (500 18TB disks/day ...) ... assuming SD quality 220 mb per hour single format and one copy.

Nitpick:

Each bigger resolution stacks on top of the previous one, so 8K60fps means storing 8K60 (webm+av1), alongside 4K60 HDR (webm), 4K60 (webm+av1), 4K (webm), 1440p60 HDR (webm), 1440p60 (webm+av1), 1440p (webm), 1080p60 HDR (webm), 1080p60 (webm+h264+av1), 1080p (webm+h264), 720p60 HDR (webm), 720p60 (webm+h264+av1), 720p (webm+h264), 480p60 HDR (webm), 480p (webm+h264+av1), 360p60 HDR (webm), 360p (webm+h264+av1), 240p60 HDR (webm), 240p (webm+h264+av1), 144p60 HDR (webm), and 144p (webm+h264+av1).

And as noted, a given resolution is often stored in multiple video formats, so there's that too.

The data needed to store a single video thus adds up quickly:

  $ youtube-dl -F 1La4QzGeaaQ | grep -o '[0-9\.]\+.iB' | sed 's/MiB/*(1024^2)/;s/GiB/*(1024^3)/;s/.*/(&)/' \
    | tr '\n' + | sed 's/\(.*\)+$/scale=3;print(\1)\/(1024^3),"G\n"\n/' | bc -ql
  8.423G

So for https://youtu.be/1La4QzGeaaQ (a random 8K video I found), storing 5 minutes 37 seconds requires about 8.5GB due to all the transcoding.

It's quite possible YouTube also keeps the original uploads around for a bit as well.

---

It incidentally proved so complicated to correctly type out the format list above manually it was actually easier to figure out the following regex:

  $ youtube-dl -F 1La4QzGeaaQ | sed -n '/video/{s/^.\{35\}//;s/ \+[0-9]\+k \+, \+/ (/;s/webm.*\|vp9.*/webm/;s/avc.*/h264/;s/av01.*/av1/;s/4320p/8K/;s/2160p/4K/;p}' \
    | sort -hr | sed ':1;H;N;${:2;s/\([^\(]\+\)(\([^\n]\+\)\n\(\1\)(/\1(\2+/;t2;s/\(\n\|$\)/), /g;s/, $//};b1'

(The above took probably twice as long to type out as the list it produced, but it was a lot more fun, and importantly verified its own correctness, which I was having a hard time doing manually. Plus, it's cool to make sed do new things - like token deduplication and newline coalescence in this case (remove the 2nd sed command to see what I mean).)

elorant• 6 years ago

A fuckload. There are 160M active domains. Say 2% of them are news sites with tens of thousands of active articles, another 2% is e-commerce sites and forums with also tens of thousands of pages, and the rest have like 50 pages on average. So you have something like 154M sites x 50 pages each x let's say 200KB per page that makes it 1.540 TB of data. And for the rest 4% let's assume 20.000 pages on average by 2MB each, that gives another 250.000 TB. And then you add sites like YouTube which could easily triple the above numbers. You'll end-up with a million TBs at least. That's like 100.000 HDDs.

catsdanxe• 6 years ago

The website that lists all possible Bitcoin private keys would take up all of those drives anyway

recuter• 6 years ago

Stackoverflow is about 10GB: https://archive.org/details/stackexchange

English Wikipedia is about 15GB: https://en.wikipedia.org/wiki/Wikipedia:Database_download#En...

Reddit is about 1TB: https://www.reddit.com/r/datasets/comments/3bxlg7/i_have_eve...

Common Crawl (May/June 2020):

"It contains 2.75 billion web pages or 255 TiB of uncompressed content". About 50TB compressed.

https://commoncrawl.org/2020/06/may-june-2020-crawl-archive-...

I think 100TB would be plenty. We're not quite there yet but I wouldn't be surprised if you could get "The Internet" on one consumer drive in a few years. This would be a cool product.

rasz• 6 years ago

Something tells me that 50TB doesnt include video media. There are dozens of chinese pirate streaming sites with more movies and TV than whole of netflix and all its competitors. Then there is YT.

sphix0r• 6 years ago

Kinda in the phase of buying disks for a $500-ish~ raid 5 NAS setup. The whole recent SMR debacle keeps me a bit careful on buying those disks before extensively checking out all the specs.

Will news like this cause retail price drops anytime soon? Other advice is welcome too :-)

dannyw4• 6 years ago

No. The HDD market is a duopoly.

You’ll find binned versions of these drives in 3 months sold at 40% off in EasyStores and Elements though.

It’s how they get rid of parts that work fine but don’t pass the specs for these enterprise drives.

dirtyid2• 6 years ago

I remember reading a comment here recently that a few terabytes is enough to store personal history of all web browsing for life (minus media). Trying to recall if that's just an observation or an actual application somewhere. So much content is disappearing, starting to seem like prudent practice.

ysleepy• 6 years ago

I've got almost three years of web history with full text locally in compressed 10gib, if you go for images and video it will get bigger though.

dirtyid2• 6 years ago

How are you doing it? A few terabytes is a small price to pay for all the esoteric personal interests lost over time.

muxator• 6 years ago

Seconded! How did you do it?

floatboth• 6 years ago

minus media

Well that's not fun. :D

I'm filling up my 8TB SMR drive with archives of my favorite YouTube channels..

sillysaurusx2• 6 years ago

In my experience, buying the “latest and greatest” hard drive is a recipe for disaster. They should put a sticker on the box that says “warning: will likely die within one year.”

Maybe I’ve just gotten unlucky over the years though. Can anyone else confirm/deny that the latest hard drives tend to be unreliable?

(Cue sampling bias...)

EDIT: The article is about an enterprise HD, but I meant to ask a general question about consumer-grade drives. Presumably enterprise drives are much more reliable, because reasons. (It would be fascinating to know those reasons though!)

mceachen7• 6 years ago

If you look at the CrashPlan drive reports, that doesn't seem to be their experience (and I don't know of a current, larger, public hard drive reliability report).

It seems that newer models trend with priors from that brand, with some outliers.

sillysaurusx2• 6 years ago

Thank you for pointing out BackBlaze!

https://www.backblaze.com/b2/hard-drive-test-data.html

https://www.backblaze.com/blog/backblaze-hard-drive-stats-q1...

mceachen7• 6 years ago

Oops. Pretend I said BackBlaze instead of CrashPlan.

hyperman12• 6 years ago

One reason: If you send a flush command to a consumer grade hd, it returns before the flush is done. This makes it look nicer on older HD benchmarking software. Enterprise hds return when the flush is complete, so databases acknowledge a commit only when it was actually succesfull.

pkaye• 6 years ago

If you send a flush command to a consumer grade hd, it returns before the flush is done.

So essentially its a no-op?

polskibus• 6 years ago

How can I find out which drives exactly have synchronous flush and which don't? Is it a widely publicized fact?

Hello712• 6 years ago

source? this sounds like an urban myth. there's no way that every single consumer grade hard drive "returns before the flush is done", given that: 1. consumer grade is not well-defined, 2. there are many different "flush commands", and it does not depend on whether a drive is "consumer grade", and 3. if this were the case, then filesystem and database corruption would be rampant, which it is eminently not.

it would be impossible to write a remotely reliable high-performance database or filesystem if flushing were useless. professionals doing formal analyses like https://www.sqlite.org/psow.html would be a complete waste of time, because everyone would be just praying anyways.

james412• 6 years ago

I think your wires may be a little crossed. This is unlikely to have ever been true for magnetic drives, but there is a quite significant grain of truth (no pun intended) to it for SSD.

Consumer SSDs may indeed return an early indication that a flush is complete, but whether they are lying depends heavily on your notion of persistence:

- the SSD may return done once the data is buffered in RAM

- or once it has reached a temporary log

- or once it has reached a 'permanent' location on flash

- or once it has recorded the mapping of data<->location to some separate directory

- or once it has updated some separate root pointer pointing to the latest version of the log/mapping structures in some separate special storage

I don't believe any SSD waits until the last step above before returning done. Separately, the SSD:

- may have no electrical power outage protection whatsoever

- may have supercapacitors fitted

- the supercap may guarantee full durability of every acknowledged flush

- or only guarantee atomicity up to some prior acknowledged flush

- or only guarantee that the drive has /some/ data on it in /some/ state on the next reboot

SSDs are insanely complex

squarefoot• 6 years ago

Maybe I’ve just gotten unlucky over the years though.

It's not you, it was WD (and apparently other vendors too).

https://blocksandfiles.com/2020/04/14/wd-red-nas-drives-shin...

pilsetnieks2• 6 years ago

Those aren't the "latest and greatest," and they didn't die per se, they're just slow for some workloads (some of those workloads that they're marketed for.)

bserge3• 6 years ago

I stick with things that have been out for a few years. Consumers seem to be just unwilling beta testers of new hardware, later revisions are more reliable. Applies to everything from computers to cars to software heh

boring_twenties2• 6 years ago

I had a few drives die in rapid succession, but then I realized I wasn't cooling the system properly. That was around 12 years ago, and I'm still using the drives I bought after that today.

Maybe you have a similar problem?

m463• 6 years ago

I've had friends who worked at hard drive manufacturers.

They said companies can very accurately know the lifetime of a drive the day it is manufactured just from the vibration profile.

enterprise > consumer drives.

It also depends on the "customer" buying the drives - certain system manufacturers will get better drives.

I think it's like restaurant owners getting better cuts of meat than you can get.

nwallin2• 6 years ago

The performance specified in the article lists only its sustained transfer rate, with no mention of sustained read speed, sustained write speed, random 4k reads, random 4k writes, or any other useful performance metric. The caches are freaking enormous- 512MB.

Am I to understand that these are SMB disks? The article does not mention SMB at all, which is surprising to me given WD's recent history.

wmf• 6 years ago

Hard drives are hard drives. They can do around 100 IOPS.

These are CMR not SMR as the article mentions.

nwallin2• 6 years ago

...Ah. I see now why the article didn't mention SMB. Because that's not a thing. Oops.

truth_seeker• 6 years ago

Continuing Hard Disk Drive Innovation from Western Digital - https://www.youtube.com/watch?v=jPoS94iVC0c

jacquesm2• 6 years ago

My personal recipe for buying drives: wait until Backblaze has run them for a year and then pick the winner. That gives a pretty safe strategy. I also buy them in small lots to avoid having the same firmware and number of hours on them, and order 3 (cold) spares so I don't have to wait for an order to come in in case a drive dies. Call me paranoid.

akulbe• 6 years ago

I think it pays to be paranoid with hard drives. They all fail, it's just a matter of when.

ASIDE: I wish someone had a calculator for which drives were a good buy right now. They get to a certain point and the bank-for-the-buck ratio is for good up to only a certain size, then the prices jump.

AnssiH• 6 years ago

Something like https://diskprices.com, maybe?

jedberg4• 6 years ago

My recipe is "Go to Costco, buy the biggest backup drive they have, shuck the drive and put it in the NAS. When one dies, repeat."

I end up with a great variety of drives and firmwares that way.

I'm still waiting on that 2TB drive to die so I can put in a bigger drive and expand the array!

ed25519FUUU• 6 years ago

I’ve heard that many external drives can’t be shucked any longer, as they don’t include proper sata connectors. Have you encountered that?

jolmg3• 6 years ago

I did that once, and found that even when connected directly via SATA, the drive seemed limited to USB3 speeds...

I guess that's cost reduction? Like, why make a faster drive when it's meant to be accessed only via USB3?

It was one of those Seagate Slim 2TB external drives.

bufferoverflow3• 6 years ago

Costco currently has an 8TB drive for $140, while Amazon has one for $135 with free delivery. And it's a Seagate drive, they don't do so well in Backblaze reports.

dannyw4• 6 years ago

Drives get swapped behind the scenes often. Waiting a year isn't reliable.

You should just assume every hard drive has a conservative annualized failure rate of 4%, jumping to 20% after 3 years, and plan your RAID arrays based on that.

jl6• 6 years ago

What is this thing Anandtech have started doing with the back button, somehow inserting an extra ad page in between getting back to HN?

nik_0_0• 6 years ago

Awful. “Check this out before you leave”

And it includes their articles plus sponsored ads about casinos for me.

manigandham• 6 years ago

They’re called bounce or exit ads. Meant to catch people from leaving too quickly, but extremely annoying and filled with clickbait crap and articles from 8 years ago.

mleonhard• 6 years ago

The drives are filled with helium. What happens when the helium leaks? I assume the drives will slow down or fail instantly.

iPhones use MEMS-based clocks that fail when exposed to helium. Are any servers, NAS devices, or laptops also susceptible to disruption from helium?

cptskippy• 6 years ago

The pressure differential between the drive and the atmosphere isn't going to be that great, if any, so it's pretty unlikely that a leak will occur.

If the Helium were to leak out then the drive would probably fail much more quickly than it otherwise would have but there isn't nearly enough helium to affect any MEMS devices.

Also consider that these drives are going to be in a system with forced air cooling so any Helium that might escape will be quickly vented.

Additionally, MEMS devices don't fail but rather behave differently while under the influence of helium.

himinlomax• 6 years ago

As cptskippy said, the helium is at atmospheric pressure, so a leak would not fail catastrophically. Helium would more or less slowly diffuse. If the leak is small enough, helium could get out but larger molecules would have a harder time. Thus the pressure inside the enclosure would get slightly lower than atmosphere and leaking would slow down.

The iPhone MEMS failed when a rather large amount of pressurized, cryogenic helium was leaked during an MRI machine servicing. The quantity of unpressurized helium in a tiny HDD enclosure is minuscule, and thus the slow diffusion is unlikely to register above background levels, even at very close range.

Western Digital's 16TB and 18TB drives: EAMR HDDs enter the retail channel

Related Stories