I've just reverted my ReadyNAS firmware back to a version that doesn't properly deal with 4k sectors, so I'm likely going to pick up a Seagate 2TB drive instead of the WD. That will be tomorrow at the earliest, so I suppose I still have a day to change my mind, depending on how my ReadyNAS testing goes...

Ugh. After a massive slew of drive failures here and with a buddy last year, I now specifically recommend AGAINST any Seagate drive purchases. For an outdated NAS, just install the compatibility jumper onto the back of a WD EARS drive. Or get a Samsung.

Cheers

Fully agreeing with Mark here Bruno: please, save yourself the agony and DON'T buy Seagate! If you want to stay away from 4K drives completely, there's also still the WD20EADS drive series which isn't 4K, but also only has 32MB of cache...

EADS drives aren't available locally - they haven't been for a long time as I also looked a few months ago. The EARS is not on the ReadyNAS compatibility list.

I have 4 Seagate 1.5TB drives in the NAS right now, all with updated firmware. The one with the most reallocated sectors is up to 18 - it started at 0 on December 8th of 2009.

I also have 4 Seagate drives in my PVR which have been doing well. 2x 500GB, a 300GB and a 250GB.

The problem reports on the Seagate drives (in general) do worry me and that's why I was looking at something else. I'd jump on the Samsung if I could buy it locally. Perhaps I will take the chance with the WD and apply the patch to change the parking time-out to 30 seconds as well as the jumper to operate in legacy mode.

I ran it on an Acer Aspire R3610, which uses an Intel Atom 330, which is a Diamondville, so yes: it supports x64.

If you can't find a Samsung drive anywhere, I'd take the WD option -with or without jumper hack- any day of the week over Seagate.

Heh, I love how around every 1 to 2 years this board has a hard drive debate. In all the previous threads it seems like there's been one person who, for example, swears by Western Digital because they had 10 Seagate drives fail on them in two years, and then it seems like there's someone with the exact opposite experience. I want to say that Rob R hated Western Digital or something...

I understand why they didn't, but I wish Google had released the results of their hard drive failure analysis. I think they indicated a fairly clear winner.

Anyway, these days I find myself going almost completely by price, as long as it isn't my main drive. Anything valuable is backed up, and anything that isn't is just a minor annoyance when/if it's lost. However, given a choice I'd probably go with Western Digital. They make good products, and their external drives are incredible. I have two 500GB USB drives that are silent from about 1ft.

I've never personally owned any WD drives. I've only ever used them when I worked at ATI where I saw an inordinate number of them go tits up in various Macs.

For personal use I've primarily used Seagate since the early 90's (all SCSI in the 90's) with a bunch of Maxtor (PATA) spread in there as well. All my SATA drives have been Seagate except for a single Maxtor. I've had 2 Maxtors and 2 Seagate drives die. One of them (can't remember which) was my fault for connecting a power plug from an external case upside down.

Eh. Everybody has different opinions about the various drive manufacturers. I know people that swear by Seagate. Others that are WD only etc...

I personally won't touch a Maxtor drive because I've had so many fail but I've been okay with Seagate. Maxtor was bought out or merged with Seagate so no idea whether I'll trust them now.

IBM/Hitachi were okay and I did even have a few of the Deathstar drives but I've not bought any of their drives recently.

If I had to buy a drive right now then I'd probably get a WD.

My first hard drives were Connor hard drives. They gave me a lot of grief since two of three drives died after some time of use. Then Seagate bought Connor and I figured, let's give them a try. Bad idea. Another dead drive, but to be honest, a couple of good drives as well (medalist series). So then I tried Maxtor. Worst idea yet. EVERY Maxtor drive I've ever owned failed on me. Then Seagate bought Maxtor. And I started having problems with Seagate (there must be some red line here). Back then, WD was always a bit more expensive than the other brands and I was a poor student, so I still thought with my wallet. But then I thought, more expensive or not, I'll buy a few of those Caviar drives. Best choice yet. I've never had ANY WD drive fail on me. Even though they were a bit more expensive than their competitors (and usually still are, of only a couple of euro!), I thought they and my data were well worth it.

I've stayed loyal to the brand up until Samsung started making hard drives and with the F1 drive created a harddisk that was hard to ignore: it was faster, ran cooler and was cheaper than any competing brand. AND they turned out to be excellent drives! I now own 6 of them and none have failed me yet, even though they gave me some problems at first (but that turned out to be the fault of some crappy external enclosure chipset). So since then I've bought Samsung, mainly because they offer so much value for money. But I still love WD too and for this reason I've bought a Velociraptor drive to use as my primary HD last year.

That's my story... I'm sure others have different experiences, but many of my friends have the same. Maxtor is the absolute worst. Then Seagate. I've never touched Hitachi since their drives already had problems in the days they were still called IBM. (mainly heat issues and that irrirating creeping sound every so many minutes when the hard drives parked their heads - it seems they never got their act together from what I've read)

I'm not saying WD drives or Samsung cannot fail. Of course they can, they are mechanical objects. I AM saying though that with these brands, you'll have the best chance of actually getting your money's worth with drives which will serve you great until YOU decide to replace them with a newer model.

I think it's reasonably obvious that a large portion of this is anecdotal (though not necessarily irrelevant) data. There are, though, a number of noted widespread failures, like the Deathstars and the recent WD Green head-parking issue.

The thing that I've noticed, though, is that no one manufacturer is always the best. I suspect that there's enough cutting-edge engineering going on that it's always a crapshoot when a new model comes out. Part of the problem is that new models seem to be more and more frequent, so that bugs tend not to get worked out over time, and there's not a long enough in-market time for there to be any significant feedback. That is, by the time that the drives die, they're already well into a new model.

At this point, I have absolutely no idea what to get. I've had nasty failures from each of the notable manufacturers, and the couple of new manufacturers scare me. (I can't seem to find references now, but I could swear I saw some boxes not too long ago with brand names I'd never heard of.)

I think the best bet is to run RAID and have, if at all possible, different drives as members, in order to try and prevent coincident failures.

I don't entirely agree with that. Hard drives development seems to have stagnated in the last couple of years. Up until a few years, drive sizes were rapidly increasing. It seemed like every so many months, the capacity doubled. That came to a halt once the 1TB barrier was reached. Then everything slowed down considerably. It took a lot of time before the 1.5 TB drives were released, and now 2TB has been the maximum size for quite some time (2 years?). So what are HD manufacturers doing with that 'extra time'? Most of them don't spend it on SSD development, that's for sure. Of all the classical hard drive manufacturers, only Western Digital has now got an SSD (which is not particularly good). Samsung has them too, but that's to be expected from a company which also produces the flash ram chips needed to create such an SSD.
I also can't believe the're now spending MORE time on ironing out bugs, because, as you say yourself, there tend to be more bugs plaguing released hardware these days than fewer. So what are they directing their efforts at?

Data density is still increasing -- look at 2.5in drives, which have carried on getting bigger -- but the manufacturers have for some reason been using that headroom to build drives the same capacity with fewer platters, rather than larger capacity. Presumably part of the reason is the thing with traditional partition tables only dealing with drives up to 2TB, but that can't be the whole story (people have had RAID arrays >2TB usable for a while), and you'd still think there'd be a market for capacious (if slow) drives for media servers and so on.

Peter

Indeed. Maybe Quantum did have a good point with their Bigfoot series then. Just kidding.

You know, I almost mentioned those but decided against it. They got a lot of ridicule for being slow, which they unarguably were, but they had more capacity than anything else on the market. A Bigfoot made in today's technology would be 10-20TB.

Peter

You're assuming that they can just scale their existing technology forever. The HD manufacturers aren't just going oh well people have enough storage we'll just sit here and wait.

The jumps in areal density are because they're introducing new technology to boost the number of bits you can pack onto a platter. The last big jump in storage capacity was because of Perpendicular Magnetic Recording (PMR) being added which significantly reduced the size of each bit. They're getting close to the limits of this technology for HDs and the next generation technology just isn't ready yet. Read this article.

Heat dissipation and power consumption is getting to be a big thing these days. The storage appliance people are having trouble with packing in large numbers of drives into enclosures now because of this. As areal density gets higher, the effects of vibration are also getting to be pretty significant. There is video of somebody at Sun with monitoring software running on a SAN that could actually detect somebody making a loud noise nearby from the increase in access time due to vibration.


If PC manufacturers ever move away from the ye olde PC BIOS and on to something like EFI with GPT support then 2TB+ drives wouldn't be a problem.

I'm sure there's still a market for such drives. Their access times were indeed horrible, but that wouldn't matter too much if they would be used for storage/backup of large amounts of data. They would be ideal in video/backup servers. I'm pretty sure I would be interested in buying one of those.

Interesting article! Not very encouraging though... I'm hoping the new technologies will be available soon!

In my fantasies I visualize data storage in a piece of crystal the size of a sugar cube, containing petabytes* of data, addressed/accessed by three tiny lasers at mutual right angles. Data transfer speed limited by speed of light considerations.

tanstaafl.

*Megabyte = 2^20 bytes
*Gigabyte = 2^30 bytes
*Terabyte = 2^40 bytes
*Petabyte = 2^50 bytes
*Exabyte. = 2^60 bytes

Whatever happened to "Pixie Dust?" I've been trying to Google it but all I find are articles about the future of it (mostly dated 2001-2003), and nothing about what happened. It doesn't appear to be even mentioned on Wikipedia (aside from Tinker Bell).

"Pixie dust" = Antiferromagnetically-Coupled media

It has already been implemented.

Ok, I picked up the WD20EARS and now I'm making a book CD to be able to run wdidle3 (I have version 1.5 - hopefully it's suitable)

Mark, for use in a RAID (ReadyNAS' modified 5/6), do you recommend 30 seconds, some other value (in multiples of 30 seconds only) or completely disabling the idle timer?

I think anything from 20-30 seconds should be good enough.

I use 30 seconds in the MythTV box.

Here in the UK I go by "who offers an advance-replacement warranty". Currently only Samsung AFAIK.

I figure that most of the errors are a few bad blocks and usually part of a RAID - I want the replacement drive around before I pull the failing disk and do a ddrescue on it.

Now I'm wrestling with abysmal performance on my ReadyNAS. I'm talking about writes of less than 10MB/s (often between 1 and 4 MB/s) and reads from 6 to 30MB/s depending on what's being read...

Dell has the WD20EARS for 100 today only. Sorry for late notice.

http://accessories.us.dell.com/sna/produ...5-404255-148326

Are you sure you actually got the head parking issue resolved? Have you checked your LCC (Load Cycle Count) to see if it is a reasonable number and the WIDLE3 program did what it was supposed to do?

tanstaafl.

I used the wdidle3 program to report back the idle time after setting it. I set it to 30 seconds, then read it back, then set it to 2 minutes and it reports that back as well. I was just going to disable it completely, but thought 2 mins was a good cmpromise between that and 30 seconds. Even if it were set to 8 seconds, the drive would never be idle that long on the copy tests I've been doing - 73GB copy is going to spin all the disks for a long while.

I'm pretty confident this is some issue with the latest ReadyNAS firmware which is the first one to support 4k sector drives.

As a point of reference, the same 73GB copy between internal SATA drives averaged out to 56MB/s. I would expect time to the NAS of at least 50-60MB/s, with the internal drives and PC being the bottle neck.

If this issue gets resolved then I can team the two network ports on the PC and the two ports on the NAS to see how much more speed I can squeeze out. BUt first things first. Other people are getting unteamed speeds upwards of 80MB/s writing to the same NAS.