Hard Drives: Commentary and Tips
Mar. 11, 2009
Mar. 12, 2009
May. 21, 2009
Jun. 30, 2009
Jul. 1, 2009
Jul. 17, 2009
Jul. 18, 2009
Jul. 20, 2009
Jul. 22, 2009
Aug. 21, 2009
Aug. 22, 2009
Aug. 24, 2009
Aug. 25, 2009
Sep. 8, 2009
Nov. 3, 2009
Nov. 4, 2009
Nov. 5, 2009
Nov. 10, 2009
Nov. 28, 2009
Dec. 14, 2009
Dec. 17, 2009
Dec. 18, 2009
Dec. 21, 2009
Aug. 26, 2010
Mar. 8, 2011
Mar. 9, 2011
Mar. 10, 2011
Mar. 11, 2011
I have now done FireWire and USB2.0 speed testing of a Western Digital WD1001FALS 1TB SATA hard drive (7200rpm, 32mb cache) hooked to an older MacBook Pro (2.16Ghz Core 2 Duo, 3GB RAM) using two different adapters.
The first adapter is a cheap USB->SATA from NewEgg ($18, PPA 1561 3 in 1 Drive Adapter) lets me hook up IDE and SATA to USB (bare drive). Xbench disk test on this adapter was 41.53 overall (only tested once, and multiple testing usually shows some variance).
The real test was using the USB/FireWire->SATA adapter I got from CoolDrives. This unit lets my plug up USB2.0 or FireWire to the drive, and compare.
Overall USB2.0 scores were 39.80-41.25. Overall FireWire400 scores were 45.72-46.04. Pretty close overall.
The largest differences were 4K Uncached Sequential Read (7.6MB FW, 5.4MB USB) and 4K Uncached Random Write (2.27MB FW, 1.85MB USB).
For 256K size blocks, read and write (sequential or random) were neck and neck with USB even taking a tiny lead a few times.
Overall, surprising results to me. I was expecting USB to be much slower than FireWire. Perhaps the laptop is a bottleneck in these tests.
I have also read that USB is more CPU intensive, which would be bad if using USB vs FW external drives for multi-track audio or video editing. (Even with high speed FW drives, I often find GarageBand and Final Cut unable to keep up with some of my projects.)
I am looking for a tool to test CPU usage during these tests and see if it makes much of a difference.
If it ends up being "fast enough" going USB, the iStarUSA 2-drive "trayless" USB/eSATA enclosure I got for $85 may make an economical way to expand storage. (I'm hoping I can use this USB/FW->SATA adapter with a larger bay SATA enclosure, but won't know if this works until my eSATA->SATA cable arrives in a few days.)
Hope this is interesting to someone else going through the same challenges as me...
[See also the follow-up below plus Benchmarking: AJA System Test -MacInTouch]
I am noticing that formatting (0-write) the 1TB drive using USB reports "11 hours" in Disk Utility, and "9 hours" for FireWire...
I've had such good luck - such reliability - using SoftRAID for striped and mirrored volumes this last decade, that I use it for non-RAID volumes, too. That and Disk Warrior. They give me confidence in my offsite backups.
[One big advantage of using SoftRAID for non-RAID drives is that SoftRAID will track I/O errors on the disk, which you can check using the SoftRAID application. (And you can do this with your boot disk, too.) -Ric Ford]
Walter Dufresne lauds the reliability of SoftRAID, and Ric Ford adds that SoftRAID will track I/O errors. To get the latter, one must select SoftRAID/Preferences/Monitor and click the boxes beside "Disk has an error" and "Disk has a recoverable error". Then, if there is an error, a dialog will be displayed.
[There are several error-logging options among the preferences, including those, and you can also log errors without also requesting pop-up dialog notification when they occur. Another feature of SoftRAID is its ability to optimize a volume for your specific needs, from "server" to "workstation" or "digital audio" or "digital video". -Ric Ford]
I use SoftRAID to make a Mirror Raid 1 out of my boot disk and two external eSATA drives in a SeriTek SATA-2EN2 dual bay enclosure that is connected to my late-2005 G5 PowerMac by a SeriTek SATA-2SE2-E PCI-express host adapter. I had crashing problems with Retrospect 6.1.230 backups to my mirror raid when SoftRAID 3.6.7 was used. However, setting the Preferences/Driver of a SoftRAID v 4 beta demo to "Submit reads and writes one at a time" solved the problem and has caused no crashes.
This SeriTek host adapter also allows booting from the eSATA drives and checking SMART with SMART Utility.
Many MacInTouch readers are probably already aware of this, but one thing to note about G-Technology drives is that they are from a company called Fabrik, which not long ago bought SimpleTech to be their consumer-level external drive line. More recently, Fabrik was bought by Hitachi, which happens to manufacture hard drive mechanisms.
I'm not saying that's a good thing or a bad thing for future buyers of G-Tech or SimpleTech external drives. Just something to be aware of.
System Profiler (v. 10.5.7 here) has a handy feature under Hardware>USB. For any attached USB device selected, it will show both the amount of current available (usually 500mA) and the amount required by the device. I don't know if (or how well) this works when a hub is attached. Anyone?
I hope this is helpful.
[Under Mac OS X 10.4.11, "About this Mac" seems to show hub information clearly, as well as details about the devices (including "sub-hubs") connected to the hub. -MacInTouch]
Regarding disconnects of the external USB drive on my MacBook Air: System Profiler shows "current required" but I am not sure it makes much sense:
Iomega drive = 2 mA
Ethernet adapter = 250 mA
Griffin SmartShare hub itself = 100 mA
I tried refreshing the view in System Profiler when Time Machine was
backing up but the current did not change. In any case, disconnects of
the USB drive seem to be happening every time the Ethernet adapter and
drive are both connected. It had been working fine for more than a
month. I don't know if this is due to a recent OS update or a hardware
problem. In any case, I thought the MacBook Air USB port had more
current available to accommodate the SuperDrive.
With hard drive prices the way they are, my policy of 'replacing any hard drive older than 365 days' is strictly enforced.
It's either $100 now or $3,000 down the road (excluding down time) when the drive fails. Why risk it? Those in the production environment should bill the hard drive into EACH job. As the job is finished, take away the hard drives and let the clients have them.
Paul Huang may be exposing himself to more risk, not less, by replacing
his hard drive every year. Hard drive failures, like many other things,
show a very definite bathtub curve; the failure rate is high early on from
infant mortality, then low for a while, then high again as wear takes its
toll. Google wrote a widely-publicized paper a couple years ago titled
"Failure Trends in a Large Disk Drive Population" that illustrates this;
the 3-month annualized failure rate they give is about 3%, while the
1-year annualized failure rate is 1.7%. His best bet if he really insists
on replacing drives preemptively would be to keep the drive in service at
a moderate load (use it in your kid's computer? ;-)) for 3 to 6 months
before moving anything mission-critical to it. At that point, you have a
year or two before the failure rates start increasing again. (Google notes
that the data presented in their paper for failure rates past one year are
skewed by having different mixes of drive models at any particular age.)
Personally, I've done well so far by only replacing my hard drives when the need for more capacity dictates, which probably aligns moderately well with the point in time where drive failure rates start to increase.
$3,000 down the road (excluding down time) when the drive fails.
While I'm all in favour of having a plan to deal with drives before they fail, I wonder if Paul could clarify where he gets his "either $100 now or $3000 down the road" price from.
The only way I can see that is if you're assuming you'll have to do data recovery...it would seem more prudent to do backups.
Personally, rotating drives every year seems excessive even for high demand production environments. In such an environment, I'd probably consider two years.
I *usually* plan on replacing my system drive on my home machine every three years, though I've yet to assess how Time Machine impacts that schedule for me... I might consider it safer to wait now (but I hate the downtime.)
There seems to be a bit of talk of replacing healthy drives before they fail. That's an aggressive approach, for sure, but one that I've personally not found necessary. I use Hitachi/Seagate/Samsung1TB drives, and over the years, the failure rate has been extremely low -- so low that my primary servers are still running Level 0 arrays with confidence -- and, obviously, that's as risky as it gets. In the last 10 years, I've had a couple of arrays go down, but because of the automated Retrospect backup to a redundant drive set, no data has yet been lost-- just the time needed to replace a drive and restore the data from the backup array.
On machines that run single disks (as opposed to RAID), I just make sure the backup drive is on board and Retrospect is scripted for the automatic sync. Should a drive fail, I then replace it. Drives are cheap, but time for some is not.
Its worth mentioning that in the last 15 years I've come to possess more
confidence in a 1 year drive than a new one. By far, over hundreds of
drive installs, I've had more drives fail in the first 3 months than
after that. In my circle, if a drive lasts a year, it will usually
outlast my system.
The storage requirements of personal data and archives are changing more rapidly than the storage manufacturer market can keep up. With the advent of digital music, photos and video; with financial services companies encouraging customers to take paperless statement delivery and to potentially store those documents on their local computers; and with the general collection of data over a lifespan, the integrity of this non-enterprise level data is becoming as important as corporate data And yet, all of this important personal data is being stored on increasingly failure-prone spinning disk, which we are supposed to believe is reliable and trustworthy.
It is not sufficient anymore for storage manufacturers to provide vast quantities of storage in a fast, cheap and small package. This faster, cheaper, smaller, higher density mission that drive makers have been on for as long as I can remember is no longer beneficial to them and not to the average end user. Right now, all faster, cheaper, smaller and higher density has gotten us is commodity quality spinning disk storage devices. From maker to maker, there is no way to tell how long a device is going to last in service or how reliable we can expect it to be. It is only after a series of drives starts to fail in a predictable fashion, or when a manufacturer gets a lot of bad press for having a broad range of failures that we know who to avoid or who to do business with. Even then it is a total gamble. This is of course after many have been affected by data loss and time reconstructing data and computers. Hard drive failure should be the exception, not the rule, especially when computer hardware itself has become so reliable that computers are only replaced when they are technically obsolete.
Drive makers have tried to salve over this issue of persistent drive failure by offering generous drive warranties and streamlined drive return policies and procedures. Western Digital has in my opinion one of the easiest to use warranty claim systems available. But it doesn't negate the fact that when I'm doing a warranty claim a drive has died in service, usually prematurely, and as such a cascade of extra labor, lost time and data are all part of that warranty claim. By the time I get to the point where I have to replace a hard drive and then rebuild the associated computer, the fact that I'm getting a warranty exchange doesn't really make my life much easier and doesn't make me much happier. More than anything else, the ease of getting a drive exchange is nothing more than a way for WD, or whatever manufacturer I'm dealing with, to avoid having supremely angry customers in their camp, not just annoyed and unconvinced customers.
This just illustrates how wrong the storage makers have it. They are dealing with their quality problems on the back side, not the front side. Just about every other manufacturer of technical products or hard goods understands the expectation of consumers that items will stand the test of time and operate as expected. Storage makers seem to be making a product where they are basically ok with full digit percentage failures in a product that shouldn't see failure rates that high. The problem is that I don't see any one or multiple storage makers choosing to focus on reliable drives and supremely low failure rates. Why? Because that would increase the cost of the drive beyond what the market has been conditioned to expect. We are conditioned to purchase $75-100 terabyte drives, but their reliability is spotty. Would a $200 or $300 terabyte drive be more reliable? I'm inclined to believe so. If we do some basic math, a $200 terabyte drive that lasts 6 years in service costs $33/year. On the other hand a $100 drive that lasts 2 years is $50/year. Now, the $100 drive might be replaced under warranty, and by virtue of that you are given a new lease on life, but what about the data loss, time lost, etc. Isn't that worth something?
This gets me to my final point. I don't see drive manufacturers increasing prices and making sure quality and longevity are built into the product's design. What this means is that computer makers and drive enclosure makers are going to have to step up to the plate to compensate for the drive manufacturers lack of quality assurance. Average desktop computers are going to have to come standard with RAID arrays and multiple disks installed to compensate for the general unreliability of today's disks. Will the next generation of the iMac contain double disk bays and an onboard RAID controller? Probably not. Maybe Apple should think about that. In the end the extra cost of designing hardware to compensate for bargain basement disks is absorbed by the consumer anyway. So much for the benefit to end users of inexpensive storage.
Sure, manufacturers and industry experts can beg and plead with end
users to do backups, but in the end, the most reliable backup is an
automatically created redundant disk, where recovery from a dead drive
is minimal to non-existent, not a periodic backup that requires a lot of
work to recover from, even if no data is lost. I view the role of
recurring periodic backups as best for archiving materials and being
able to restore individual files that become corrupted or lost or
mis-edited. Restoring a computer from a periodic backup is possible, but
not fun or easy.
Here's how I do it:
As soon as the system is set up, I clone a startup volume and put it away. Except the iMac, all other current Macs hard drive can be easily replaced within five minutes.
Backup is a given, not an alternative - sounded like someone was insinuating that I am trying to eliminate backups by replacing the drives. It is certainly not so.
I also remember that Google did a study on heat and found that the correlation between heat and hard drive failure is low. Well, [I don't agree with that].
As for the 'reduced failure rate over time', I have yet to find that to
be the case in my own situations. Besides, I have a backup scheme and
direct link to file servers, so I wouldn't have time to worry about
that. Oh, by the way, as each project is done, the drive is removed and
goes with that project. A new drive is installed. It's a standard
operating procedure now.
2.5" drives are not as durable. Although there are mechanical reasons, the biggest reason is movement during operation. Sudden twists and turns could cause a lot of strain in the bearings.
I will touch upon another topic: hard drive replacement. While desktop drives are sitting still at the same place and not moved around, notebook drives are subject to vibration and sudden moments. This probably causes the 2.5" to have a higher failure rate over time. From the hundreds of 2.5" drives I have replaced, more drives have failed outside of the one-year period. So my statistics completely refutes a reader's report that if a drive does not fail in the first few months, it is likely going to out-last the computer itself.
By the way, my entire household (except a Core 2 Duo iMac) runs on 2.5"
(Mac portables and Mac Mini) drives. If I were using a desktop, I may
not stick to the 365-day rule.
I keep a DVD disk image file (4.7 gigs) for all the important files that need to be backed up.
You know what they are - some candidates:
... anything vital to your organization
Back them up with Disk Utility burn, and mount them to verify the information is readable
I've only *really needed* these disks once, but it was a lifesaver.
I've found that a little self-discipline beats Time Machine any day.
Ref: Jerome Parmentier and mfgrs should make more reliable drives....
They do already. Virtually every manufacturer has a line of "industrial" drives rated at 5 years or more. They are better balanced, use better bearings; better motors and so on.
Visit their sites and you'll find them.
The topic of disk drives reliability comes up periodically on MacInTouch. I have been working in the disk drive industry for many years, and I have commented on the topic in the past on these pages. Jerome Parmentier's post is eloquent, but mostly wrong. He puts the blame on drive makers - guilty, in his opinion, of selling products with sub-par reliability.
Disk drives do fail occasionally. That's a fact of life. But it is also true that disk drive reliability has been steadily improving in the past several years. This is based on inside knowledge. As I have said before, manufacturers are understandably reluctant to share reliability data.
It is the same tremendous economic pressure that drives (pun intended) the price of disk drives down that makes it an absolute must for the survival of a manufacturer to keep a low failure level. The cost of replacing a single hard drive easily negates the meager profit on the sale of many units. He who has unreliable products will be severely and brutally punished in the marketplace by its OEM customers. The disk drive market is a rare example of a free market that really works, in this case to the advantage of the customers, who are getting a terrific deal, and to the detriment of the drive makers, who are struggling in a fierce competition and hardly make a profit.
Jerome's question "would a $200-300 terabyte drive be more reliable?" is then really only a rhetorical one, rendered meaningless by the economic realities, as such a drive will never be built. And if it were, I doubt that it would be significantly more reliable than today's products.
What is then the best approach for the end user to cope with hard drive
failures? It is to render such failures harmless by having full,
automated backups. Apple has to be commended for introducing Time
Machine. With it, you can sleep sound knowing that a disk drive failure
will be dealt with by replacing a drive and restoring your data from the
backup. If you use Mac OS X but not Time Machine, you have really only
yourself to blame.
Jerome Parmentier makes good points about drive reliability and the race to the bottom.
I think that drive failure is an inherent problem of spinning platters and moving read/write heads. Think about the duty cycle: A 7200 RPM drive running around the clock rotates on its bearings 3.7 BILLION times per year.
For comparison, the wheel bearings of a car that drives 15,000 miles per year turns 12-14 million times per year (depending on the wheel size). And its pretty common to replace some failing wheel bearings after a decade or so.
Even the read/write arm sees heavy duty: If the drive seeks just a hundred times a minute on average, that's over 52 million seeks. (That's very conservative. Especially if you turn your computer off and on every day. Or even just launch applications and browse the web.)
With the sort of duty we expect of them, it's amazing that hard drives last at all!
There are more reliable drives made, but they are not cheap. They're intended for servers and are very high-margin because, well, they can get it -- Mr Parmentier aptly describes the alternative.
A terabyte drive for a desktop PC or Mac costs around $100 today. A 600 GB Seagate Cheetah 15k.6 drive intended for servers costs $500. Most people buy the cheap, big drive, not the expensive, small drive. (And what a way we've come -- the first *giga*byte drive was sold in 1980 for $40,000 and weighed 550 lbs.)
Perhaps the way out of this lies in solid state drives -- with no moving parts, they shouldn't be subject to the same wear that spinning drives do. (Consider, by the way, that the fundamental design of a hard drive hasn't changed since IBM created the Winchester drive in 1973; we've just been refining it ever since.)
Of course, we might end up in the same state of SSD's as we are with camera storage and CD/DVD media today -- a race to the bottom for most, and a few expensive options if you know where to look and are willing to pay a premium.
But maybe, just maybe, the price difference for these future drives might be merely twice as much for a good device (like flash media for cameras is today) instead of the factor of ten difference we currently suffer with.
Until then ... I'm not willing to pay SAS prices for drives like that Cheetah (not to mention the $900 Apple RAID card required to use them in a Mac Pro), so instead I just try to buy drives with a long warranty, put a backup system in place, and hope for the best.
2.5" drives are not as durable. Although there are mechanical reasons, the biggest reason is movement during operation. Sudden twists and turns could cause a lot of strain in the bearings.
As the adoption of 2.5" drives has increased, it's no longer truly valid to make a blanket statement like "2.5" drives are not as durable." Many modern 2.5" drives are as good or better than their 3.5" cousins, especially if you're comparing a 3.5" drive that's a year or two old.
"Sudden twists and turns" can happen with any drive really, and the
impact of movement is not exclusive to 2.5" drives. For most people
using a laptop sitting on a desk while in use there aren't going to be
any sort of "sudden twists and turns" when it's in use. When the drive
is inactive and heads are parked, any movement shouldn't cause damage to
Just a quick comment on Jerome Parmentier's post about hard drive reliability:
"Now, the $100 drive might be replaced under warranty, and by virtue of that you are given a new lease on life..."
Not so much, when you remember a warranty replacement is a *reconditioned* (read: used) drive. More like a *used* lease on life!
To follow up on Colleen Thompson's comments on blown Firewire ports:
Several years ago when Firewire ports were blowing left and right, I
developed a protocol, based, if I recall correctly, on one of James
Wiebe's (of Wiebetech) white papers, to help clients reduce this damage. I
called it PLUGOFF! (Putting it in caps seems to help people remember it.)
PLUGOFF! means that one should *never* touch a Firewire cable plug to
disconnect it from, or plug it into, a computer or other device unless the
external source of the cable, say a dismounted Firewire hard drive, is
powered *off*. If you have a chain of Firewire devices, dismount them (if
necessary), turn them off one at a time starting at the end away from the
computer, then disconnect the cables in the same order. Mount them in
reverse order. So far, neither I nor any of my clients have lost a
Firewire port using this PLUGOFF! protocol. YMMV of course.
A MacInTouch Reader wrote:
"It's hugely more likely that the drive's Firewire port itself (or the plug on the cable you're using) has died, especially if it sees heavy use (plugging/unplugging)."
Definitely try a new cable, if you haven't already done so. I had a flaky device return back to full health as a result of swapping out a bad cable.
If your port is indeed the culprit, it might still be a firmware issue and not dead hardware. Many have reported success resetting dead FW ports by zapping PRAM, or an OpenFirmware parameter-reset or doing a PMU/SMU reset. The procedure is different for different model Macs, but it's something you may want to try.
I've also seen reports about dead FW ports being a software issue. If you have a FW-based bootable device, try booting from it and see what happens. If that works, then the port is fine. Re-apply the latest Combo updater from Apple to see if that replaces the flaky software driver with a good copy.
Back when blown FireWire ports were a hot topic and far-east research
showed that connecting a powered unit to the Mac's port sent a voltage
spike which would blow the non-resettable logic board fuse, Wiebetech
began selling a port-isolating cable that only passed data. I bought one
as the safest way to deal with FireWire at the time. Another step I took
to encourage data transfer reliability was to purchase a supply of ferrite
beads (toroidal chokes) from an online electronics parts supply house.
These go on the data cables I use for USB & FireWire backups to neutralize
any stray RF in the cables during operation. So far, so good.
I see that there is some concern with drivers being available for RAID devices.
Just wanted to note that the very reliable SoftRAID product drivers are included in all builds of OS X since 10.4; one does not need to worry about an external source.
Messing with electronics is humbling no matter how long you've been doing it. I have been doing it since 1968 starting with Heathkit radios and now list "messing with Macs" as a hobby even thoug I use one at work to run my business as well.
Last weekend I had a scary two-hour experience with my Firewire 800 port on my MBP 17 unibody. I plugged the cable in, started up the drive and -- nothing. Could not get the drive to mount. I plugged the cable into another drive. No joy. I ran every diagnostic I could think of. No joy. I tried a drive with the USB interface. It worked. I tried a drive with the eSATA adapter card. It worked. I was stumped, frustrated and concerned. Finally I unplugged all the cables, shut down the machine, reset the SMU, rebooted into Safe Boot mode and then plugged the FW 800 cable in.
Oops. Whoa. It went in upside down. That was alarming in a totally different way. Now I really had some adrenaline flowing. I got a Maglite flashlight to see if there was something broken or maybe some lint in there.
Nope. Everything was perfect -- except the dummy at the controls.
Despite having correctly used the FW 800 port dozens if not hundreds of
times, in this instance I had somehow been plugging the Firewire 800
cable into the Mini DisplayPort. The FW 800 connector is a near perfect,
snug mechanical fit, and will go in right side up or down. The ports are
right next to each other. Fortunately no wrong contacts were ever mated,
so no damage, except to my pride.
[And if you're fumbling around in the back of the computer without a good view or good light, the Ethernet ports are also a similar size and shape, but you really don't want to stick a FireWire 800 cable in one.... -Ric Ford]
Sterett Prevost mentioned FireWire isolation cables.
Actually, WiebeTech didn't have a port isolation cables, someone else
did (FWDepot, I think). WiebeTech did include spike protection on their
FireWire drives so such a cable was actually not needed. Unfortunately,
the isolating cables do not pass bus power, so any attached hard drive
must be plugged in.
Thanks to the reminder from an alert reader, the FireWire port isolation
cable I got long ago (?) really did come from FireWire Depot, and yes, the
drive has to be powered up with its own path to AC power as that cable
only passes data.
So the Iomega eGo drive case is not aluminum after all. I decided to open the other half. I noticed that there was rust on the areas where the plastic case has molded windows to accommodate for the Oxford 934 chip and other ICs. I have posted pictures on Amazon.com's product page.
While the package clearly indicates 'aluminum', but it's just a metal shell glued on the outside. It's not even aluminum, but steel instead. On certain foreign Iomega sites, they even say 'full steel'.
I wonder if the contract specified aluminum, but the contract manufacturer switched to the less expensive ungalvanized iron or steel.
I have read the articles on HW-Encryption Drives. Since I investigated
this a lot, Macs still lack the ATA Security Feature because they did not
implement this BIOS-Feature in EFI.
But there is a solution:
Seagate offers "Enterprise" FDE-Drives. I tested these with SecureDoc for Windows and they have a special area for a preboot environment.
WINMAGIC (www.winmagic.com) offers SecureDoc for Macs. SecureDoc can administrate FDE Drives on OS X and so instantiate the PBA which then unlocks the drive's encryption hardware to boot OS X!
Unfortunately Winmagic only sells 10 Licenses at a time, so one would
have to spend 1.500 Euro for 10 Licenses from which probably only one or
two would be needed.
Otherworld Computing (Macsales.com) has a new blog entry about the hard drives in the new iMacs that concerns me: Proprietary cable can put the brakes on upgrading Late '09 iMacs
As we were getting information together for the new iMac instructional videos, we came across a little tidbit that, apparently, hasn't been covered anywhere else: Apple has switched the iMac's method of hard drive temperature sensing. They've gone from an external sensor that attached to the outside surface of the drive to a connector that seems to use the drive's internal sensors. [...]
That means, in order to upgrade the internal drive, you need to have a connector cable that's compatible with the brand of drive that you're installing... and that's an Apple service part not generally available to the end user.
If the iMac isn't getting the temperature data from that sensor, it puts the fans into wind tunnel mode. So, for the time being, any upgrades will need to be with drives from the same manufacturer.
OWC includes a list of drives they have confirmed work in the newest model iMacs (known as "Late 2009" in Apple parlance).
Dell started using proprietaryinterface cables years ago that included temperature settings for things such as system fans. Not on all their machines, but many. It forced the user to buy replacement parts from Dell or their authorized provider.
It was a bad move then, it's a bad move now.
The temperature connectors vary from one manufacturer to another, thus Apple has to match the manufacturer's connector.
There's been some discussion of SMART Utility here recently; anyone
who's thinking about trying it might like to know it's currently on
special at 20% off via the One
Finger Discount promo (scroll down to Volitans Software). Also, an
article goes into some detail on how to use it.
The 2.5" drives sold packaged in external cases aren't necessarily
standard. Some may have direct USB instead of SATA connectors and some may
be taller than even 12.5mm drives:
Attention: external USB disks from various constructors, continued
Say, another holiday shopping tip ... I noticed that Other World Computing has a version of the Voyager Q SATA drive dock without FireWire, but with USB + eSATA, for just $33. That's less than what I paid for my BlacX dock last year, and the Voyager has better design and usability, imho.
If you're looking for a drive dock, here's the link...
Voyager Q Drive Dock
The Thermaltake BlacX costs a few bucks more at Amazon but is eligible
for Prime or free shipping: Thermaltake
Constantin von Wentzel
I see Robert Mohns is advocating the Voyager Q drive over at OWC, a site sponsor. While the combination of ESATA and USB 2.0 seems like a "good enough" solution for most of us, I cannot recommend this dock when used in conjunction with the OWC-branded ESATA Expresscard/34 at this time.
Tech support was able to replicate the problem at their end but has not found a solution as to why this combination cannot zero out drives using disk utility. It works on USB 2.0 (though slowly at about 25MB/s) and large file transfers get up to 70MB/s on ESATA. To me, zeroing drives is the primary reason to buy this sort of dock, they are not case substitutes, as cases cost relatively little, the drives in them can be upgraded over time, etc. Here, an investment in a quality drive with multiple interfaces (such as the Aluminum series from OWC) pays back with years of trouble-free operation.
At a $30 price point, the dock represents a nice solution to zero drives that you'd rather donate/sell than destroy. Yet, until the folk at OWC figure out why the Voyager dock and Expresscard combination do not work, I'd spend the extra cash for a quad-interface Voyager dock instead. At least that dock offers FW800 as a backup interface.
Lastly, it's too bad that OWC will charge a restocking fee for any returned item, even ones that OWC acknowledges are faulty. I offered to pay the difference to buy a working solution, i.e. the quad-interface dock, and got a big letter of the terms of sale in return. Oh well, sign of the times, I suppose.
[Just to be clear: Robert Mohns has no financial ties with the company other than as a customer, but he did write a review of the Voyager Q, so he's especially familiar with that product. I also have purchased quite a few products from the company at full retail price, as a customer (mostly hard drives). -Ric Ford]
Constantin von Wentzel
Hey Ric and Robert,
I in no way wanted to imply that reviews here were somehow influenced by OWC being a site sponsor. I apologize if my letter somehow implied that, it wasn't the intent. I too have been a happy customer of OWC for a number of years now, and the last time I had an RMA, it wasn't a big deal, they simply exchanged non-working merchandise for working merchandise. Since they confirmed that my present issue seems to affect all products that they sell, a simple product-for-product exchange won't do the trick though.
Hence my suggestion to return the two affected products and pay the difference to buy the quad-interface dock instead... a suggestion that was acceptable to them, as long as I pay the restocking fees for both products. This stance may or may not be allowable under consumer protection statutes, credit card agreements, etc. but I find it puzzling to treat a customer this way when a product has been shown to be defective. That said, there is still hope that technical support will manage to find a solution. I'll let you know if they do.
Re the problem of Von Wentzel:
I, too, have been a customer of OWC for a very, very long time and have always found them to have the highest level of customer service. It is unusual for them to react in this manner.
I suggest that a manager be contacted. If the product is defective it should be returned without a penalty.
It would seem that OWC's excellent reputation may be one of the reasons that Mr. Von Wentzel is willing to purchase another product. Let's hope OWC keeps it that way.
Steven James May
Really sorry to see this. Back in the early PowerBook days I felt that the best hard disks were those Apple sourced from IBM. No rigorous testing or survey, it was really just a feeling I got over time from working with dozens, perhaps hundreds, of PowerBooks -- a kind of intuition I've learned to pay attention to. So when I needed an aftermarket disk I looked for IBM.
Then IBM's disk division was sold to Hitachi, and I transferred my "brand loyalty" to Hitachi. And I've noticed that I seldom see complaints here or elsewhere about Hitachi disks, while Western Digital and Seagate, the two biggies, seem to provide endless problems for users.
Apple has used a lot of Toshiba and Fujitsu disks in recent years, but I've gotten a bad taste about them after seeing so many fail early.
So now Hitachi disks (if they keep the brand, which I suppose is doubtful) will actually be Western Digital disks, a make I've distrusted for 20 years or more, since first encountering their lousy products in the late 80s. *Sigh*
So what's left? As a Mac user, I like to figure out what's best and then stick with it. The only one I don't have an impression about is Samsung; guess maybe I'll try them next.
I agree with Fred Moore's sentiments and Andrew Main's experiences completely. I too went from IBM to Hitachi because of problems with WD and Seagate drives. Guess it's time to buy that big Hitachi drive while it's still in stock.
It's been my experience that every hard drive manufacturer has had a run of iffy drives. If its not a component, its the firmware. These problems get fixed either through updates or returns. Placing your trust, forever, in a single manufacturer is just not workable.
Andrew Main commented
"Back in the early PowerBook days I felt that the best hard disks were those Apple sourced from IBM. No rigorous testing or survey, it was really just a feeling I got over time from working with dozens, perhaps hundreds, of PowerBooks -- a kind of intuition I've learned to pay attention to. So when I needed an aftermarket disk I looked for IBM.
Then IBM's disk division was sold to Hitachi, and I transferred my "brand loyalty" to Hitachi. And I've noticed that I seldom see complaints here or elsewhere about Hitachi disks, while Western Digital and Seagate, the two biggies, seem to provide endless problems for users."
You must have missed the "Deathstar" iteration of IBM's Deskstar line of drives. Failure was so rampant it led to a class action suit, 18th place on PC Worlds 2006 "Worst Tech Products of All Time" list and became commonly know as the "Deathstar" line.
Andrew Main wrote:
"Then IBM's disk division was sold to Hitachi, and I transferred my "brand loyalty" to Hitachi. And I've noticed that I seldom see complaints here or elsewhere about Hitachi disks, while Western Digital and Seagate, the two biggies, seem to provide endless problems for users."
You see more complaints about WD and Seagate for two reasons:
1. They sell a lot more drives.
2. They sell the highest performance drives, therefore, they are the choice of power users, who are also some of the most vocal users.
"Apple has used a lot of Toshiba and Fujitsu disks in recent years, but I've gotten a bad taste about them after seeing so many fail early."
Can't say anything about the Toshiba drives, but the Fujitsu drives have been of comparable reliability to the Hitachi, WD, and Seagate.
"So now Hitachi disks (if they keep the brand, which I suppose is doubtful) will actually be Western Digital disks, a make I've distrusted for 20 years or more, since first encountering their lousy products in the late 80s."
If you haven't used a WD drive in 20 years, you haven't used a WD drive. That's like comparing Seagate drives of the last 10 years to the Seagate drives from the 80's (e.g. before they purchased Imprimis and Connor Peripherals). Quality has changed dramatically in the last 20 years. In the first half of the 90's, IBM 5.25" HDs were some of the worst in the industry, and in the late 90's or early 2000's their "deskstar" drives earned the nickname "deathstar".
Of all the existing HD "brands", Fujitsu is the only one that (to my knowledge) isn't known for a high failure rate at some point in their history. On the other hand, Fujitsu also isn't known for making high performance drives (at least they haven't been since the early 90's). In particular, the Fujitsu laptop drives are known for mediocre performance, low power, and good reliability.
Andrew Main wrote regarding Western Digital's acquisition of Hitachi's disk storage division (HGST):
"Really sorry to see this."
and wrote about his experiences with and impressions of other brands.
Well, it's been a while since IBM/HGST had a major lemon (75GXP "DeathStar", anyone?), but I've had a few of them fail in my PowerBook G4 (yes, in one laptop).
My conclusion is that no one brand is perpetually the best. Every brand has phases of high and low quality.
Around the turn of the millennium, I wouldn't touch a Western Digital with a 10 foot pole, they were dropping like flies back then. Then IBM had the 75GXP fiasco. For the G3/early G4 era, I stuck with IBM/HGST for notebook drives and Maxtor for desktop.
Then Seagate bought Maxtor. For the late G4/G5 era, I used Seagate with no problems. But now, in the Intel era, I've had multiple new Seagates fail. Meanwhile, Western Digital seems to have not only caught up, but exceeded the others.
All in all, it's hard to say. Not only does quality vary by brand, it also varies significantly by model. (Google's famous drive reliability survey showed that some models were particularly prone to failure, while other models from the same vendor were very reliable.)
For now, I'm using Western Digital and Hitachi, phasing out my Seagates before they croak. Samsung seems to be fine, but WD makes higher-performing drives -- and nobody has a product to compete with the Velociraptor.
What will I be using in 5 years? Who knows. That's assuming there's more than one drive maker by then?
I'd like to provide a counter-anecdotal experience to Andrew Main regarding Hitachi hard drives. My experience has been that the IBM/Hitachi drives are in the lower 50% in terms of quality and reliability.
Even the relevant Wikipedia article mentions the infamous "Deathstar" hard drives:
My company was the victim of those drives. We bought 50 IBM POWER workstations with the aforementioned drives. Within 60 days, *all* of the drives failed - not 60%, not 80% - we're talking a 100% failure rate. They were of course replaced under warranty, but the loss in downtime and excess IT support costs were not reimbursed.
This was one of those times when I wished IBM had used Western Digital drives. I've been using WD drives for over 10 years and haven't had a drive go south on me yet.
My experience with external WD drives has not been good. I have two black MyBooks which have failed - one the controller went and the drive is now in a third party MISCO supplied enclosure. The other has the click of death. The drive in the third party case is running very sweetly so far after 1 month.
Also a 2TB Studio drive has now been replaced twice in 3 months under warranty.
Andrew Main said:
"Back in the early PowerBook days I felt that the best hard disks were those Apple sourced from IBM."
Perhaps that was true, but even IBM had their clunkers. The failure rate on the IBM DeskStar drives was so high that people nicknamed them the "DeathStar".
Broadly speaking, IBM's drive division was sold to Hitachi because of the
Hitachi improved quality to be sure, but a bad batch of disks can happen anywhere.
Western Digital and Seagate have vastly greater numbers of drives in
stalled, making it fairly natural that you were hear a much higher
number of complaints.
RE: WD acquistion of HitachiGST
So now Hitachi disks (if they keep the brand, which I suppose is doubtful) will actually be Western Digital disks,
There is still likely going to be split post merger between "enterprise" and "consumer" drives. Since Hitachi seems to have deeper penetration in the "enterprise drive" space, it seems likely those will continue (at least as model names/numbers) and the WD offerings in that space will fade away over time.
For example, a search for "SED" or "TGG Opal" on wdc.com website turns up empty. On the other hand Hitachi hits:
Likewise Hitachi has "enterprise" SSD drives and WD doesn't. Hitachi has to have something WD doesn't or else it is awfully dubious to spend $4B to acquire it.
So buying "low end" Hitachi drives may fade away but if really looking for very low defect, higher quality drives they should stay around.
Note also that the person currently running the Hitachi subsidiary is going to be in charge of making drives in the new company. Perhaps an executive "payoff" (buyin for merger prospect), but perhaps indicative going to bring over some of the IBM/Hitachi lineage in new company.
Hitachi drives' reliability:
Since Apple went ATA with their PowerBook products, I have set up more than 1000 units of Mac portables. I don't recall the IBM/Hitachi as particularly reliable compared to the other brands.
I do, however, see a high correlation between how callously the users
handle their portables and the failure rate - regardless of brand.
Personally I prefer WD to any other brand of hard drive. I will not willingly use a Seagate on the other hand. I have been very lucky over the years with hard drives with only one complete failure. On the other hand, I have had no end of trouble with enclosures from Seagate and Iomega. I have a Hitachi enclosure that drives me nuts for being both bright and deciding to wake up (and be bright) at really inconvenient times. My WD drives have so far just worked. I also have a few LaCie drives that I like but given a choice, its WD for me.
As others have said, WD today is not WD from 20 years ago and their sheer volume of sales results in more 1%ers. Add to that their very successful MyBook line brought external storage to a lot of non-technical users - users who fill a drive to the last MB. Users who unplug without unmounting, pile things around the vents and subject drives to shocks. You can hardly blame WD for some of the failures that get reported. Visiting a non-technical friend one day I was appalled to see a coffee cup sitting on the top (vent side) of a MyBook. There were coffee stains all over the drive housing. *When* it fails, I am sure he will be one of the people talking about crappy WD drives.
WD also has some of the best Mac support of any drive maker. You can buy drives preformatted for Mac and most of their external drives are compatible with Time Machine. MyPassport Studio drives feature Firewire 400/800.
Hitachi does not sell a single FireWire drive or preformatted Mac drives - something Seagate, Iomega, LaCie and WD all do. Iomega even offers a drive that matches the Mac Mini and includes a USB hub. Of the major makers, Hitachi seems to me the most Mac hostile.
As for SSD replacing conventional drives - maybe for use in laptops but not for serious storage any time soon. SSDs run between $1.50 and 3.00 per GB. Conventional drives run $.03 to $.10 a GB. I have no problem shooting 8GB of pictures in a day and have shot as much as 32GB for an event. Because I back everything up, an 8GB shooting day "cost" me $.96 (8GB x$.06 per GB = $.48 x 2 = $.96). That same shoot would cost me $24 using a low end SSD. Until SSDs drop below $.20 a GB they will be uneconomical for mass storage. By the time they do, I expect conventional drives to be about $.50 a TB.
Here we go again. Lots of people talking about how each and every brand of drive is trash, and lots more pointing out how they've never had problems.
The fact is that there is no single brand that has always been perfect. All of the big-name manufacturers are reliable, but all have released models with high failure rates.
The solution isn't to declare entire brands forever off-limits, but to pay attention to the news so you can avoid those specific models that are known to have problems. (And avoid any brand-new models for a few months, so allow time for any problems to get reported in the news.)
Re: Kenyon Kopp
Speaking of DeathStars: my Power Mac G4 (2002) came with one within the affected serial # range. I quickly replaced it with a larger drive and put it in a cheap CompUSA USB case to use as temporary backup, file transfer, et al. It's still working fine, and it's still relatively fast.
Robert Sholl writes:
"Hitachi does not sell a single FireWire drive or preformatted Mac drives - something Seagate, Iomega, LaCie and WD all do. Iomega even offers a drive that matches the Mac Mini and includes a USB hub. Of the major makers, Hitachi seems to me the most Mac hostile."
No, it just appears that way at first glance. Hitachi owns G-Tech, which makes some very high quality external drives with FW800 and 400 as well as other interfaces. We use them extensively in our all-Mac business.
Certainly hard drives are complex mechanisms. The good ones are masterpieces of high-tech hardware and software engineering. Thus, any hard drive manufacturer can screw up, cut corners, or just have bad luck which results in a spate of low-quality products with higher than average failure rates.
With respect to Hitachi and the IBM Deathstar, notice that the class action suit was filed *2 years before* Hitachi acquired IBM's hard drive operations in 2003 (ref: the Wikipedia article cited yesterday). Hitachi then fixed the problem, though it did take some time. My experience, that of my clients, and that of others I've read about, is that the reliability of both Hitachi desktop and laptop hard drives is above average to well above average. I will admit that since many hard drives I specify are intended for backups, I usually recommend Hitachi's 'enterprise' grade drives for maximum reliability. They are, of course, more expensive but almost bullet-proof workhorses. Unfortunately, my low-cost supplier, ZipZoomFly, is currently out of them and almost all other Hitachi drives (a harbinger?) or I'd buy a pallet load.
Regarding Robert Sholl's comment that
Hitachi does not sell a single FireWire drive or preformatted Mac drives...
There is the G-Drive series by Hitachi:
Their "Professional External Hard Drive" is HFS+ formatted and includes Firewire 800 (as well as eSATA and USB 2.0).
I've seen the 2 TB model of the G-Drive "Professional External Hard Drive" at my local Best Buy store.
Neil Maller wrote
No, it just appears that way at first glance. Hitachi owns G-Tech, which makes some very high quality external drives with FW800 and 400 as well as other interfaces. We use them extensively in our all-Mac business.
I sit corrected :)
Sneaky little hobbitses. Wicked, tricksy, false!
G-Tech makes a very nice line of drives but I had no idea they were owned by Hitachi.
Still like WD best.
Amen to David Charlap's comments.
Some of the complaints heard in this thread are not about the HDAs themselves but about other components - enclosures, power supplies, logic boards - that are part of a Hard Drive manufacturers retail, off-the-shelf total solution. Seems unfair to brand a whole HDA drive as unreliable when the culprit might be something else.
Had a WD MyBook that wouldn't mount. Problem turned out to be a flaky USB port, not the HDA. Since it was out of warranty I removed the drive and was pleasantly surprised to find WD had used an enterprise-class RE-series HDA, not a consumer-level HDA.
Popped that HDA into a new OWC enclosure and all is working fine.
For external storage I prefer buying the best HDA one can find and assembling it into an enclosure I can trust based on experiences and recommendations from all of you who post on MacInTouch.
One thing that is often overlooked (& which we've discussed over at AMUG) is the required startup power to boot your hard drive. This does vary by brand, with Hitachi taking among the least, and Seagate taking 50% more. (22 watts vs 35 watts.) If you have a big power supply (such as those inside your Mac, or inside an external enclosure) this isn't a very big issue.
However, if you're using an external enclosure with that's powered by a brick or wall-wart, the situation isn't the same... particularly if you're using an enclosure that supports multiple drives.
Those power supplies often won't last much beyond a year or so because of the strain of trying to power up all those drives. (Manufacturers know this and some better units will offer staggered starts, but many don't.)
Solutions? Buy a better enclosure; buy a spare wart/brick; don't fully populate the enclosure; or use Hitachi (or others that can boot more efficiently.)
Here is a fascinating synopsis of a study on hard drive reliability done
The Durability Standard
Here's a website with lots of hard drive & SSD info:
From the reliability study Hitachi stands out with the fewest failures due to manufacturing defect (none), all Hitachi failures in the study came from "accidents" (drops etc.) Digging in a bit more, from what I can recall the Hitachi spindle was more sturdily constructed/anchored, where other drive spindles might start failing earlier due to lubrication issues or issues with wobble of the spindle due to poor manufacturing.
For what it's worth, I just replaced a Fujitsu 120GB with the "click of death" in my sister-in-law's Macbook. I've also had very bad experiences with LaCie externals (older ones, probably due to the power supply failure, I'm too lazy to tear them apart to check.) I've never experienced a Seagate 2.5 or 3.5 inch failure, although the laptop drives seem to run hotter and are noisier than others. Never had a Hitachi, Samsung or WD failure either, I hope I'm not jinxing myself!-)) I do use a variety of brands in my backup strategy to avoid the "one bad batch" problem.
All the best in your hard drive explorations!