Technical > PC Hardware Discussion/Reviews > Hard Drive Quality Nosedive in October?

Hard Drive Quality Nosedive in October?

I think people who are unhappy more likely to leave feedback.

In any case, if you're worried about quality, you can do a 'burn-in' for a day or two to deal with infant mortality, then run them in RAID 5 or 6. (unless they're WD, I guess )

I find it hard to believe that quality would drop simultaneously for all makers, although there aren't that many. In addition to Seagate and WD, there's also Hitachi and Samsung with 1 GB drives.
--
Put a stick lip on a pig!

reply to disconnected
Vocal minority.

reply to disconnected

I;ve checked Hitachi drives too since my last post and there's a rash of very negative reviews starting in October, too.

I doubt it's a vocal minority, unless they just discovered NewEgg reviews last month, because reviews prior to Oct were overwhelmingly positive. From Oct onward, WD, Seagate and Hitachi have a flood of DOA reports for the reviews.

Here's that issue that a review posted and WD tech support's explanation of 'deep recovery' mode on the drive:

»www.newegg.com/Product/ProductRe···22136284

Don't Use in RAID

Pros: 1TB - If they work they perform fine but nothing spectacular over other 1TB drives.

Cons: These drives run hot but that seems to be the same for most 1TB drives. I'd recommend some direct cooling.
Western Digital does not support running these drives in a RAID configuration. I bought 5 of these drives. 1 was DOA (likely due to shipping as the bubble wrap was popped in areas) and one other will fall out of a RAID set every 2 days. According to WD it is because the drive enters a deep recovery cycle and the RAID controller times it out. Why it needs to enter a deep recovery cycle so often is a question they won't answer. Western Digital wants you to pay more for their RAID edition drives which don't enter the deep recovery cycle. I've never has this problem with other HD manufacturers (before these drives I used mainly Hitachi and Seagate with hardware RAID controllers). I use Areca RAID controllers which have worked great for me in the past. The 3 remaining drives I have work fine but WD won't replace this drive just because it fails in a RAID array.

Other Thoughts: You would think we should be able to run HDs in a RAID array without having to spend extra for a drive designed to operate 24x7 when that is not the intended use.
Here are some of the replies from Western Digital support:

"The WD Caviar Black SATA drive is not a RAID edition drive. When creating a RAID, we recommend that you use our RAID edition drives. We cannot guarantee RAID functionality of our desktop drives for reasons listed in the article below.

Title: What is the difference between Desktop edition and RAID (Enterprise) edition hard drives?
URL: »wdc.custhelp.com/cgi-bin/wdc.cfg···31638613 "

"It is normal to experience dropouts in RAID arrays with desktop drives. I recommend returning or exchanging the drives to the place of purchase for RAID edition drives, or contact your RAID controller manufacturer to see if they can provide a RAID controller that can see drives that are in a deep recovery cycle. "

reply to disconnected
Interesting. I can't really take Newegg reviews seriously, as there's too many idiots on there who simply lack the experience to provide a decent review.

I'm in no way denying the possibility of a problem, but step back for a moment: how likely is it that *two* hard disk vendors, of competing measure, would both implement a feature that would limit its use like this? It seems very unlikely. It also seems even more unlikely they'd implement it ONLY on their 1TB drives.

EDIT: I hadn't read the link you provided. I'm going to dig through the technical data sheet on these new disks to read more about TLER. This sounds very similar to what IBM/Hitach's ADM feature was back in the day, and they removed it once it was found by users.

EDIT 2: I just finished reading the full link. The behaviour Western Digital describes for this new form of sector reallocation is normal -- but 2 minutes to complete a sector reallocation is too long, even for desktop usage. I'm not sure what the hell Western Digital is thinking by doing this. Desktops will notice this as well! Hell, even the TLER function taking 7 seconds is unacceptable.

Also, I want to point out that the FreeBSD operating system has a hard-coded 5 second timeout for ATA transactions. If the drive "locks up" for literally 120 seconds while reallocating bad sectors, this is going to cause all kinds of chaos -- even for desktop usage. I'm going to report this finding to the freebsd-hardware and freebsd-fs mailing lists, because this is simply unacceptable. Western Digital needs a slap in the face.

That said, I'll provide some evidence of my own.

I just purchased two Western Digital Caviar GP 750GB disks earlier this week. One of them, flat out, had bad sectors right out of the box -- starting at LBA ~500 (very close to the inner edge of the platters). The drive makes no clicking noises, it just simply has bad sectors. The OS witnessed these LBA errors, which means the drive internally had exhausted its list of spare sectors.

The other drive appears to function, but SMART statistics show one (successfully) reallocated sector. I need to write zeros to the entire disk (every block/sector) to see if there are any others which the drive hasn't detected yet.

What disappoints me is that the bad sectors are supposed to be detected at the factory, and marked as part of the physical defect list (which isn't managable by end-users). This means the testing/verification process in QA has changed and they're not doing a thorough job, or that these disks are being assembled in a fab where dust is somehow seeping into the disks *after* testing.

FYI, both of these disks will be used in a ZFS RAIDZ (think RAID-5) array, and so far with one disk (see above) in use, it performs fine. Here's the story in case you want to read it.

What I'm trying to say here is that I think Newegg buyers are making crap up. :P But with regards to the "quality nosedive in October" claim, that might actually turn out to be true. So far I've only tested 2 WD Caviar GP disks (I have no interest in getting Seagate, due to increased temperatures after the 7200.10 series), and those results aren't that great, nor damning of the company.
--
Making life hard for others since 1977.
I speak for myself and not my employer/affiliates of my employer.

reply to disconnected
Probably too simple a solution for them to implement, but why can't WD just sell one type of drive and add a jumper option- jumper installed= desktop mode, jumper removed= RAID mode? This way, you could take a drive you have owned for awhile and convert it if you decided that you wanted to use it in a different application.

koitsu is right. Why on earth would you even bother taking "customer reviews" in NewEgg as any kind of accurate predictor about a product? Any fool can (and will) write a review. If this is for an "IT department", I would hope that any potential purchase would be based on a better basis than what one reads at NewEgg!

Besides which, your entire premise is flawed in one major aspect:

You said "It seems like drive quality across the board has taken a steep nosedive in the past 35 days. It's as if drive quality tracked the economy and perhaps factories are in financial trouble and just shipping drives without testing, in a cost-cutting desparate measure."

How can you say that? How do you know the manufacture date of any of the hard drives being reviwed? What you know are simply the dates of the reviews, not what manufacturing batch that these drives came from, and you are assuming that they were all built at approximately the same time, of which there is no proof either way, yes?

I do understand your concern, but reading reviews in NewEgg is not a very good way to judge.
--
...something is happening here but you don't know what it is...do you, Mr. Jones?

reply to daveinpoway

Regarding your point #1, perhaps WD could develop a way for the user to access the drive firmware and set the preferences there, instead of using physical jumpers.

As for your point #2, I agree this could cost WD some revenue. Also, it is possible that the RAID models are somehow better-built.

Feature                      Support  Enable    Value           Vendor
write cache                    yes      yes
read ahead                     yes      yes
Native Command Queuing (NCQ)   yes       -      31/0x1F
Tagged Command Queuing (TCQ)   no       no      31/0x1F
SMART                          yes      yes
microcode download             yes      yes
security                       yes      no
power management               yes      yes
advanced power management      no       no      0/0x00
automatic acoustic management  yes      no      254/0xFE        128/0x80