Anything wrong with High-capacity drives?

[Sidrack]

Pros: Good capacity....

Cons: High capacity...

Other Thoughts: High capacity drives come at the cost of volatility. When seagate released the 1.5 TB drive, I was skeptical. As a physicist, I know that capacity is limited by technology, and while miniaturization is moving quite quickly, there is still a physical capacity, regardless. with magnetic storage this dense, one slight error by the head will corrupt the file, and repeated events - all of your files.

While the storage, and other perks from reviews already posted, are pretty good, I don't see these drives having a long lifetime, and most definitely wouldn't use them in high volume and mission critical environments. I'll stick with my 1TB drives until these have been on the market for a good amount of time. Call me a skeptic, but I'd like to think that I didn't spend an obscene amount of time in school for nothing.

While this is a single comment on the 2 TB drive page (Newegg), a long time before I read what the guy said I was thinking about the same thing.

First, the Samsung 1 TB drive (HD103UJ) received a high number of complaints from people stating those drives were faulty, so many critics that I decided to purchase the 750 GB model (also the 1 TB drive was expensive that time, so two enough reasons for me). Some people stated the drives worked for months before they eventually died... which is real bad when it's something unpredictable. In that case, we can't say for sure how those things are build and if they will last the same time as the old and low-capacity models.

Well, I haven't seen any HDD dying yet (not even from old age), and mine have years of usage, but they were probably build in 2002, 2003 (Samsung 160 GB and Seagate 300 GB). So I can't say anything about younger and high-capacity drives (currently I own 2 Samsung 750 GB drives, but they are months old). And my system is used 24h/day.

And then, as you might probably know, the massive failure from Seagate 1.5 TB drives (and probably a plenty more from the same company, I believe they are known as 7200.11).

Google did a research about this:
http://storagemojo.com/2007/02...isk-failure-experience

But what I am really concerned about is not why or how the drives might die. It's the expectations we can have with these kinds of drives. Are they really so much complex, build to have a long lifetime, or the companies are investing in something that is not meant to have that capacity?

I mean, I always wondered why fast drives (with 10K or 15K RPM) have a very low capacity, and why someone haven't build a 500 GB SCSI drive for example, making the same commercially viable for most people. Then someone said to me it was impossible due to limits of technology.

Aren't we doing the same thing with normal internal drives? Do you see a resemblance with them and the tower of Pisa (the design was flawed from the beginning)? :laugh:

HDDs are old technology, perhaps a single drive was not supposed to have so much storage capacity, without those side effects, or at least a shorter (or probably risky?) lifetime.

 

[Blain]

Perpendicular recording = Much smaller bit area = Higher probability for data corruption & drive failure.
I like that Google produced some original content. :thumbsup:


Interesting podcast interviews about HDs and data recovery...

* Podnutz Episode 28: Thor Schrock
* Podnutz Episode 29: Scott Moulton

* Computer America 03/24 Hour 1: Scott Moulton
* Computer America 03/24 Hour 2: Scott Moulton

 

[RebateMonger]

The high percentage of DOA or early deaths certainly indicate that drive makers are pushing the limits of their manufacturing processes and/or physics.

 

[nyker96]

ya don't buy just one. get at least two. if they die, they take a lot of stuff with them.

 

[pjkenned]

Check out the Seagate 15k.7's and Savvio 15k.3's capacity on SAS drives these days may surprise you.

The 7200.11 issue was firmware related. I saw two dead drives from it. Since then, the other 11 have been fine, and they all had the newer firmware.

Only drives that I have non-raid 1/5/6 at this point are USB keys, my Dell Mini 9 SSD, PS3 hard drive (can't figure that out and don't see the value), and miscellaneous old disks that I use for large hot swap bay - hot swap bay/ eSATA transfers.

The big issue, as others have stated, is that if you have a single or raid 0 drive, you are exposing a ton of data to failure.

 

[cbn]

I am considering a HD upgrade in the near future. Can anyone recommend a reliable Large drive?

Are the Samsung Spinpoint F1 and Western Digital 32MB buffer drives reliable?

What about these lower Capacity 320 and 640 GB drives I see. Why are they so popular? How about the Western Digital "Green Drive"?

 

[lopri]

My answer is limited to Samsung F1 and WD Green. I had two of each (1TB drives), and tested them side by side. Both were very good performers, but Samsung F1 was faster and quieter. I was a bit surprised because I expected the Green to be slower but quieter.

Had a plan to try WD Black on top of it but I didn't, because noise was/is a bigger concern to me than performance (for these 1TB HDDs).

 

[Blain]

I read people commenting on the "noise" of the WD "Black" HDs.

Are you guys working in an anechoic chamber? :shocked:
Ambient noise around here is far louder than my WD "Black" ever thought about being.

 

[Sidrack]

Originally posted by: Blain
Perpendicular recording = Much smaller bit area = Higher probability for data corruption & drive failure.
I like that Google produced some original content. :thumbsup:


Interesting podcast interviews about HDs and data recovery...

* Podnutz Episode 28: Thor Schrock
* Podnutz Episode 29: Scott Moulton

* Computer America 03/24 Hour 1: Scott Moulton
* Computer America 03/24 Hour 2: Scott Moulton

Interesting comments quoted from the podcast 29 (you can heard in the 30 minutes/half of the MP3 file) regarding bigger drives:

Q: Is there any particular brand that you think is better than any other on hard drives?

A: I would say that all of the drives are, at least today, not great drives. They are all made, and they are all going to die, a fairly quick death, pretty much the drives that are 500 GB and larger, are dying on a daily-basis for everybody. Terabyte drives, things like that. My preference would be to buy Seagate, only because of the warranty.

So, in my opinion, at least (because the drive is going to die) I can get a replacement based on the warranty, that's my hope.

Q: But why are the bigger drives failing?

A: The thing that has happened since 2006 is that everybody is going to a perpendicular format. And what that means is before, pretty much before 2006 every drive was written in what is called longitudinal format.

And that meant the north and south poles that you used to seem on a platter when you actually think about "how my data is written, it's laying down on a platter, north and south poles, and the head reads in that direction".

Well, since 2006 perpendicular basically means it's no longer laying down on the drive, now we are going to store it up and down. It's the difference between, you know, let's instead of burying [?????] in the ground, laying down, let's bury standing up right, because they will save a lot more space.

That's exactly what's happening with the bits and data stored on the platters now. So, at least from a stand point of sensitivity, the heads are much more sensitive to the amount of space and density that they got to read the content, and everybody is saying "let's get the biggest drive we can possibly get now", so, you know, one of the consequences of that is going to be: heat, and less time to test things.

I've know people who bought 500 GB drives, bought 3 or 4 book drives, or something like that, and within a week of each other they are dying.

And it's a fairly consistent inconstant thing now that one of things that you see in for recovery are these 500 GB drives both from Seagate, Western Digital, a few other companies, but, primarly Seagate and WD, the two I see most of.

Q: I see. So, there is no real brand loyalty, they are build not great, but if you are going to put your money on something, you might wanna put on Seagate because of the warranty?

A: That's my opinion, because they tipically they back the 5-year warranty before, I am pretty sure that's what it is now, still, but, tipically some other vendors, only either add 1-year or 3-years.

It also depends if either you are buying in the external case or not, or OEM drive, cause if you are buying a Lacie, the drives inside them are not made by them, but Maxtor, Seagate, or somebody, and tipically, they only give you a 1-year warranty, you'll have to read the box to be sure, but versus what the manufacturer might offer, so you might not get the 5-year warranty when it's inside the case.

P.S. Funny thing I said that I bought the 750 GB Samsung instead of the 1 TB drive because of the many complaints. Take a look into the Newegg page here:
http://www.newegg.com/Product/...Mark=&Page=1&Keywords=

And you will see lots of reviews from people saying things like that about the 750 GB drive:

Pros: Especially quiet now that it is dead.

Cons: Purchased at the $109 price point from Newegg.com back on 5/13/2008. Drive is dead now and will now attempt to RMA.

Other Thoughts: DO NOT BUY THIS DRIVE OR ANY OTHER MADE BY SAMSUNG UNTIL THEY HAVE PROVEN RELIABLE.

Cons: Bad compatibility. I was running it on XP and kept getting write failures. Switched it to an external enclosure via USB and it worked ok. I imagine because the connection was slower. Drive failed after 2 months. Lost 600GB of data which was always destined to be backed up tomorrow. I'm not sure if I even want to try to RMA it because of all the horror stories on this drive.

Pros: None!!! Should have the ability to list as ZERO EGGS!

Cons: Drive Failure After 5 Months. Take A Close Look At The Reviews. 1 day to 1 week, All Happy. 1 month to 1 Year, Huge Amount Of Failures. Just Know That Samsung Makes It Impossible To RMA!!! Thay Staste A 3 Year Warrany, But That Is Only As Good As The Company Behind It. Samsung Sucks. Thank New Egg For Posting All These Reviews, Maybee YOU WONT SUFFER LIKE ALL THE REST OF US!!! And To You Reviewers That Are Happy, JUST WAIT. YOUR TIME WILL COME! GOOD LUCK WITH YOUR RMA!!!

Other Thoughts: Waited For A Week Now, And I Still Have No Return Corrispondance With Samsung. And If You Think They Make It Easy For Returns, Your Dreamin! They Know This Drive Sucks And Dont Want To Have To Pay To Cover There "Warranty". Shop Elseware. I paid $109.00 in Aug '08. Now Its $79.00. Wonder Why?! Thanks Samsung, Lost Customer For Life On All Products. Hope To See Where You Land In This Recession.

This is really bad...

 

[taltamir]

if you double the size, you have to double the reliability per write in order to MAINTAIN the same level of reliability per drive... while in reality reliability actually goes DOWN as it is harder to remain reliable at further miniturization levels.

Drive manufacturers still CLAIM the same "one error per #Gb written" today as they did 10 years ago... assuming drive makers are telling the truth and are not exaggerating their reliability at all whatsoever, that still means your chances of encountering errors has gone way up because you are writing a lot more GB into a medium that can store a lot more GB...

This is why I use ZFS, it has data aware raid with file level checksumming... It shows that I have about 1 write error in writing every 2 months on my 5x750GB WD geenpower raid6 array (raidz2 for zfs actually). But due to the data aware + file level checksumming it is able to FIX that error from the other drives with no actual dataloss.

any other filesystem would have had silent data corruption.

 

[Rifter]

"Anything wrong with High-capacity drives?"

Yes the 2TB ones are still to expensive

 

[Sidrack]

Originally posted by: Rifterut
"Anything wrong with High-capacity drives?"

Yes the 2TB ones are still to expensive

I couldn't agree more. :laugh:

Some people said to me all those complaints were because of the "earlier production drives"...

But the 750 GB drive was not released yesterday. And even if that was the case, it's no excuse for so many defective and unreliable units. And the Seagate 7200.11 problems...

If we were talking about any other product, I would be skeptical about the few bunch of negative reviews. Because all piece of equipment is like a box of chocolates. You never know what you're gonna get.

But we are talking about a product that already have hundreds of people complaining about defective units! This isn't something we will see everyday. Isn't that scary enough for us to reconsider purchasing, spending a lot of money on these drives? Saying this is nothing but bad luck is blissful ignorance of the facts.

Take a look at this comment from the Newegg bad reviews page. Will you say the guy was reading this thread before he posted it?!!!!

Pros: NONE!

Cons: Was $130 when I bought it around March '08. Failed on October '08. Lost 600+ gigs worth of data including music and by music I mean my own music. I'm a musician and lost 2 albums worth of music sessions. Sure some people would say "durr back up your data durr". I do that from time to time IT'S JUST THAT THERE IS NO REASON OR AN EXCUSE FOR A HARD DRIVE TO FAIL AFTER 8 MONTHS OF USE especially when you buy something from a reputable brand like Samsung.

There are also serious compatibility problems with this drive and Nforce controllers. Google it around. Plenty of people cant get their computers to register this drive. Others would have to wait 10-15 extra seconds to POST every reboot.

Other Thoughts: I've been using a Western Digital 80 gig SATA drive as my main drive for 3 years now. I've dropped it twice. I've installed, re-installed so many different OS on it. I've formatted and reformatted it dozens of times. And it still works. SMART tells me fitness is 99% and performance is 100%!!! I have another similar WD drive but IDE form and 120 gigs that I use in a VERY old Dell computer. That drive is now at least 5 years old and it still works without any problems. I have a 5 year old 3200 RPM (!!!) Maxtor external USB drive and guess what??? NO PROBLEMS!

I still try to fix this hard drive from time to time by attempting to use an external enclosure or even SpinRite but nothing has worked. No software can possibly fix a shoddy manufacturing job.

Thanks for clicking noises and loss of important data Samsung. I really appreciate it.

And even if the guy purchased one earlier unit, how can you be sure you are not doing the same thing with all other brands? I mean, is there any massive recall happening right now for all of them, any quality control, removing those units from all stores and all that stuff? All RMAs were done without problems?

We will never be 100% safe, we will never know if we are purchasing a drive produced in 2007 that is on the back of the store just waiting for the next victim.

Of course we should always make backups of every sensitive data stored in Hard drives, no one is dumb enough to forgot the most basic thing. But like I said, this is not my primary concern. I would like to know how my money is invested.

 

[Idontcare]

Originally posted by: taltamir
if you double the size, you have to double the reliability per write in order to MAINTAIN the same level of reliability per drive... while in reality reliability actually goes DOWN as it is harder to remain reliable at further miniturization levels.

Drive manufacturers still CLAIM the same "one error per #Gb written" today as they did 10 years ago... assuming drive makers are telling the truth and are not exaggerating their reliability at all whatsoever, that still means your chances of encountering errors has gone way up because you are writing a lot more GB into a medium that can store a lot more GB...

I'm not privy to the details but my assumption has always been that this is the case because the primary reliability issue is the drive's mechanics and not the actual platter/bits/etc that go bad. This is why data recovery services can exist, the data is still on the platter, assuming catastrophic physical failure of the head did not occur.

So while the likelihood of a "1" inadvertently becoming a "0" on the platter is low (and has remained low) the likelihood of the 7.2k rpm motor dying or the armature becoming misaligned or the head crashing into a platter has remained high (little to no improvement) and its frequency of occurrence dominants the MTBF stats.

 

[taltamir]

well, actually the likelihood of a 1 becoming a 0 has MASSSIVELY increased, since the smaller it is, the more likely it is for a cosmic ray or some other background force to flip it...

And I was referring to the likelihood of the arm MISSING the mark and writing that 0 or 1 on the wrong sector, or not supplying ENOUGH voltage to its electromagnet to change it (due to PSU power fluctuation)... in REALITY the smaller you get, the worse it gets... but drive manufacturers have claimed that it remained EXACTLY THE SAME... that its the same 1 in X writes is botched. that X has not changed at all... the thing is... a 2TB drive has a lot more WRITES to handle than a 1MB drive had to.

So far I have one detected "wrong" bit every 2 months on my array, which ZFS can detect and fix . Weather its cosmic rays, PSU fluctuations, head "missing" the correct write spot... whatever, I can't really tell. I only know it is a REAL problem... I only looked into ZFS becaues i noticed silent corruption on some of my files... that i find old rar files which gave a CRC when extracting that used to work before...

 

[ElBurro]

As with any drive you can never completely rely on it. If you want to be safe make a back up no wait make 2 back ups. Even the most reliable drives will fail eventually and when it does your screwed. They are so cheap anymore that how does it hurt to buy a TB drive and if it fails you get replaced. Most of them have at least a 3 year warranty so your not out the money.
Here's what it comes down to there are 2 types of hard drives 1) the type that has failed 2) the type that will fail.

 

[Idontcare]

Originally posted by: taltamir
well, actually the likelihood of a 1 becoming a 0 has MASSSIVELY increased, since the smaller it is, the more likely it is for a cosmic ray or some other background force to flip it...

I'm not disagreeing with you (or rather, what you are saying here doesn't invalidate what I stated)...I am merely stating the obvious that a small number increased by a very large percentage can still be a small number and in reliability metrics the lifetime limiting contributors are still likely to be dominated by the reliability of the mechanical components.

Originally posted by: taltamir
And I was referring to the likelihood of the arm MISSING the mark and writing that 0 or 1 on the wrong sector, or not supplying ENOUGH voltage to its electromagnet to change it (due to PSU power fluctuation)... in REALITY the smaller you get, the worse it gets... but drive manufacturers have claimed that it remained EXACTLY THE SAME... that its the same 1 in X writes is botched. that X has not changed at all... the thing is... a 2TB drive has a lot more WRITES to handle than a 1MB drive had to.

Still not disagreeing, but if the likelihood of this occurring was 1 bit for every 1E50 bits written but now it is 1 bit for every 1E40 bits written (got worse by 10 orders of magnitude) then the drive makers can continue to claim that it is still better than 1 bit for ever 1E15 bits written (then and now). (the specs I've seen were listed as <1bit for every 1E15 written, etc, and 1in 1E40 is definitely <1 in 1E15 for example)

I get what you are saying, the realized liklihood of these kinds of errors has increased with the increasing areal density, but the hardrive guys are safe (I think...) with their stated specs as they have padded them long ago in anticipation of the continual degredation in the actual error rates.

Originally posted by: taltamir
So far I have one detected "wrong" bit every 2 months on my array, which ZFS can detect and fix . Weather its cosmic rays, PSU fluctuations, head "missing" the correct write spot... whatever, I can't really tell. I only know it is a REAL problem... I only looked into ZFS becaues i noticed silent corruption on some of my files... that i find old rar files which gave a CRC when extracting that used to work before...

That's cool, I did not know ZFS could do that. Do cosmic rays flip bits in magnetic media? I thought that was just a charge/capacitor issue which is why dram and flash are sensitive to it (as well as CCD's). I've had files die on my like you are saying, never thought much about it because I keep backups/etc on anything I care for. I'll have to checkout ZFS.

 

[FHDelux]

You can never determine when your drive will go bad to the point of "i can't get my data". If you are really worried about losing a ton of data, there are a lot of options out there. Internet backup sites, burning Blu-ray / DVDs, Raid 1 or 5. Any of those things will decrease the likely hood that you will lose your data.

I have a 1TB hard drive, and i back it up to an external hard drive once a week. But other than the time the backup hard drive is backing up, it is never plugged in (not spinning, so not decreasing life span).

Hopefully SSDs will get better and more reliable so we can all switch to those. No moving parts means they are less likely to fail. In the future i would expect SSDs to be able to tell you exactly when they are going to fail (aside from freak things like electrical failure).

 

[Idontcare]

Originally posted by: FHDelux
Hopefully SSDs will get better and more reliable so we can all switch to those. No moving parts means they are less likely to fail. In the future i would expect SSDs to be able to tell you exactly when they are going to fail (aside from freak things like electrical failure).

SSD's avoid the mechanical reliability components but they are susceptible to cosmic rays flipping bits regardless whether the drive is powered or not.

Single Event Effects in 1Gbit 90nm NAND Flash Memories under Operating Conditions

Here's a nice tutorial from Hot Chips that gives a nice overview of SER and NAND (among others):

Memory Performance Tutorial Hot Chips 16 Agenda

And lastly here is a reporting on the subject matter in technical journal format (NASA):

SEE and TID Characterization of an Advanced Commercial 2Gbit NAND

Hopefully once we get flash densities to the point that SSD's are financially viable at 1TB capacities then manufacturers will be more willing to allocate 1/9 of the physical capacity towards ECC with a controller that is capable of automatically detecting and correcting SEE's to avoid silent corruption of one's files stored on an SSD.