POST A COMMENT

46 Comments

Back to Article

  • artifex - Friday, December 09, 2005 - link

    Heat? MTBF? Warranty?
    Reply
  • grnmyeyes - Monday, December 05, 2005 - link

    1. Why use the WD4000YR (RE2) instead of the WD4000KD (SE16)? From what I gather, the RE2 is optimized for RAID configurations while the SE 16 is intended for general single-drive use.

    2. While the review states that the nForce4 motherboard used for testing supports TCQ and NCQ, it neglects to mention whether these features are enabled on all the drives tested. If other sites (storagereview.com) are to be believed, turning off TCQ/NCQ can result in a substantial performance increase for single-user applications.

    3. This last one isn't really a flaw, but it would be nice to see the Maxtor DiamondMax 11 put through this same test suite when it becomes available in the near future, since it is designed to compete with these three drives.
    Reply
  • NightCrawler - Sunday, December 04, 2005 - link

    http://storagereview.com/php/cms/cms.php?loc=news_...">http://storagereview.com/php/cms/cms.php?loc=news_...

    1. First, the WD4000YR (that is, the “Raid Edition 2”) enjoys a 24-hour factory burn-in period versus the WD4000KD (the “Special Edition 16”)’s 8-hour test. The longer validation cycle increases the chances that a drive destined to suffer an “infant mortality” of sorts is caught at the factory before it enters distribution.

    2. Next, the YR features a longer 5-year warranty more typical of enterprise-oriented drives such as WD’s own Raptor or Seagate’s Cheetah series rather than the shorter 3-year coverage that backs the KD.

    3. Finally, the RE2, living up to its moniker, ships with the firm’s “Time Limited Error Recovery” (TLER) feature enabled while the SE16 does not. This oft-confusing feature permits a drive to gracefully surrender error recovery features after an 8-second window of attempts rather than stubbornly continuing onwards with the assumption that an array’s redundancy features will cover the error and avoid data corruption. For drives that find themselves in a setup that features redundancy (anything other than RAID 0), this is a plus. For disks that don’t, it’s not.

    This leveraged technology between the RE2/SE16 and the Raptor leads WD to boldly proclaim that the WD4000YR can sustain use in a 24 hour a day, 7 day a week environment with its actuator seeking 100% of the time… the only such claim for any 7200 RPM drive on the market. Competing drives such the Seagate NL35 and Maxtor MaXLine Pro, while offering 24x7 power up, are more cautious in duty cycle (that is, seek) claims- “nearline duty loads” for the Seagate and “40%-50%” for the Maxtor.

    Worth the $20 extra to get the YR version and the 5 year warranty.


    Reply
  • bluefmc - Friday, December 02, 2005 - link

    Purav,

    The WD hard drive you tested is not recommended by WD to be used attached to a non-RAID controller. To quote the StorageReview.com article:

    "In fact, WD touts the drive's "Time Limited Error Recovery" (TLER) ability, a feature that enables better drive-controller coordination in mutually handling drive errors. The drawback, however, is that the RE2 expects to operate off of a RAID controller and is as a result not recommended for use in a standard desktop system."

    You can read about this in a WD PDF at [L=http://www.wdc.com/en/library/sata/2579-001098.pdf]http://www.wdc.com/en/lib
    ary/sata/2579-001098.pdf[/L} Basically it says that if this drive is used attached to a non-RAID controller data corruption may result.
    Reply
  • irev210 - Friday, December 02, 2005 - link

    Big woopty deal.


    Go buy a WD4000KD



    Besides, the SE16 is cheaper than the RE anyway.


    https://www.zipzoomfly.com/jsp/ProductDetail.jsp?P...">https://www.zipzoomfly.com/jsp/ProductDetail.jsp?P...

    199 at zipzzomfly with free 2nd day air. An absolute bargain!

    Reply
  • bluefmc - Friday, December 02, 2005 - link

    It's is actually a big "woopty deal." If someone buys this drive to use as a desktop drive connected to a non-RAID controller it can result in data corrution and/or data loss. This drive is intended solely for use in a RAID array connected to a RAID controller. The article made no mention of this fact, so I thought I'd point it out. Reply
  • phanna - Saturday, December 03, 2005 - link

    From the PDF you posted: "A drive under a continuous I/O load and performing its own error recovery can easily exceed 8 seconds of timeout, during which the normal desktop hard drive does not respond".

    These drives are fine to use as desktop drives. The only thing this sheet is saying is if you do have this connected to a RAID controller, and you are using it in a high load environment, you don't have to worry about it being dropped by the controller if the drive has to perform lengthy error recovery, due to its TLER technology.
    Reply
  • irev210 - Saturday, December 03, 2005 - link

    quote:

    From the PDF you posted: "A drive under a continuous I/O load and performing its own error recovery can easily exceed 8 seconds of timeout, during which the normal desktop hard drive does not respond".



    exactly!
    Reply
  • bluefmc - Saturday, December 03, 2005 - link

    Let me explain a problem that has been popping up recently with the growing popularity of SATA RAID. Most SATA drives are not made for use in RAID arrays and are made for desktop use. What this means is that they will take as long as they need to recover from an error. Normally this is a good thing. This becomes a problem when such a drive is hooked up to a RAID controller, since while fixing an error a drive is non-responsive. If a drive is non-responsive for more than approximately 8 seconds, the RAID controller assumes that the drive is dead and fails it out of the array. As a result a perfectly good drive is kicked out of the array for no reason and you now have a lengthy rebuild process to go through.

    This new drive from WD solves this problem by only trying to fix an error for 8 seconds, then giving up and passing the error along to the RAID controller. With any level of RAID with the exception of 0 you will have no data loss or corruption, and you won't needlessly fail a perfectly good drive out of the array.

    Now if this drive is not hooked up to a RAID controller and it gives up on an error in 8 seconds that may have been repairable by the drive in 10, 20, 30 seconds or more, data corruption will result. And yes these longer errors do happen with some frequency. Hence if you do not use this drive in a RAID array that has redundancy, you run a risk of losing data.

    Now I want to be clear that I'm not putting down WD or their drives. I'm running 2 WDs in my desktop now myself. I just want people to understand that this drive in not suitable for typical desktop use. This drive must be used in a RAID array of level 1 or higher.
    Reply
  • irev210 - Saturday, December 03, 2005 - link

    Yup, so i guess we can say that anandtech should have mentioned - BUY the SE16 which is CHEAPER anyway if you dont plan to run a raid, otherwise you could have unneeded problems down the line.

    Makes sense to me.


    Good info bluefmc
    Reply
  • Visual - Saturday, December 03, 2005 - link

    i cant imagine what error would hang my drive for 8 seconds :/ and if it really happened, even in recoverable error, i'd not trust that drive again anyway. so it'd be better to mark it "failed" Reply
  • Lakeshow - Saturday, December 03, 2005 - link

    Yeah I read that article on storagereview.com couple days after I got my WD4000YR and it kind of bothers me.

    Oh well, what are you gonna do? I absolutely love this drive. I can only hope this drive will live until my next voluntary upgrade.
    Reply
  • Lifted - Friday, December 02, 2005 - link

    The HD Tach screenshots say "for non-commercial or evaluation use only, see license agreement."

    Hmmmmm. Anandtech is non-commercial?
    Reply
  • Gannon - Friday, December 02, 2005 - link

    I'd like to see more tests done on drives that are at least 80% full because a lot of us pack our drives full of stuff and the performance we end up getting is when we've filled it. While these tests are good and all, I think they inflate the actual scores of how a drive is really used. No drive sits with just XP and a game or two and a couple of test files, that bias's the tests toward unrealistic use of how hard disks are used, especially big ones over 160GB. I fill my drives regularly and I have over 800GB needless to say I'm backing up stuff to DVD's just to have enough space to perform other operations. Reply
  • WileCoyote - Friday, December 02, 2005 - link

    Why does everyone want Anandtech to benchmark their current system? Run your own benchmark if you want to know the speed of your hard-drive/computer. I think the articles here are perfect - they help me decide what to purchase in the future. I don't need an article to make me feel good about what I already have. I like the current format of articles that educate me on my next purchase. Reply
  • johnsonx - Friday, December 02, 2005 - link

    time to cut down on the pron addiction
    Reply
  • Olaf van der Spek - Friday, December 02, 2005 - link

    > We were also surprised to see the WD4000YR perform so well, since it is using the 1 st generation 1.5Gb/sec interface.

    That's a joke, right?
    I hope you weren't really expecting a significant performance improvement from a faster interface (300 mbyte/s instead of 150 mbyte/s while HDD's are more near 75 mbyte/s and only during seqential access).
    Reply
  • bob661 - Friday, December 02, 2005 - link

    I think the interface increases benefit RAID performance more than singledrive performance. Reply
  • Olaf van der Spek - Friday, December 02, 2005 - link

    I don't think so, as (without involvement of port multipliers) SATA is a point to point architecture. Reply
  • yacoub - Friday, December 02, 2005 - link

    Or you can go buy a pair of 200GB Samsung SpinPoint SP2004C drives for under $100 each and have a much quieter drive setup. :) Reply
  • mostlyprudent - Friday, December 02, 2005 - link

    AT, your multitasking conclusion is interesting. You give the nod to the WD drive because it has the smallest percentage performance drop when running the multitasking scenario. However, the Seagate drive was faster in both the Zip only test and the multitasking test. Theoretically, it slowed down more when moving to the multitasking scenario, but in reality was still faster than the WD. I understand conceptually that you're looking at the effect that multitasking has on drive performance, but I wonder if this is a relevant test when comparing drives. Reply
  • Anton74 - Friday, December 02, 2005 - link

    This may be slightly off topic, but for me personally (and I think a lot of folks), it'd be much more interesting to see reviews of drives that I might actually buy, which are at the moment 300GB or less - these larger drives are just too expensive per GB.

    I'd be particularly interested in the 160GB 7200.9, which to my knowledge still hasn't been tested by anyone - and it's the drive with the highest platter density at the moment, by far! But there are at least a handful reviews of the 500GB version already (which was praised with "mouthwatering benchmarks" a good month ago, but now "is just not up to par to recommend"...).

    While I'm at it, how are the (mostly) slower 3Gb/s SATA drives more "future-proof"? If they're slower than the WD today despite their faster interface, they'll be slower than the WD 2 years from now, me thinks. Or are they expected to improve with age? :-P

    Don't get me wrong here though, I'm glad you guys are spreading all this review goodness around (you were the first to review the Asus A8N-VM CSM for example, two thumbs up).
    Reply
  • bob661 - Friday, December 02, 2005 - link

    quote:

    these larger drives are just too expensive per GB.
    Too expensive?! LOL! How cheap do you want them to be? Hard drives are DAMN cheap now. One of the cheapest components of the computer.
    Reply
  • LoneWolf15 - Friday, December 02, 2005 - link

    quote:

    Too expensive?! LOL! How cheap do you want them to be? Hard drives are DAMN cheap now. One of the cheapest components of the computer.

    What he means is, that buying a 250GB drive is FAR cheaper on a cost-per-gigabyte basis. As drive size increases past a certain point in the current market, vendors charge more per GB. When you think about the fact that a Hitachi 250GB SATA2 hard disk is currently around $110, and a 500GB model is $395, why in the world would you buy the larger drive when two 250GB models cost so much less? Most of us enthusiasts have the case space, and if you really want a single large partition that badly, you can use Win2k/XP's Dynamic Drive capabilities or the JBOD feature of your RAID controller (note: I wouldn't do this personally due to data integrity issues if one drive fails, I'd just have two partitions).

    He's not whinging that hard disks are expensive; just that there's no point in buying a pricey large HDD, when two medium HDD's that amount to the same size can be had for a lot less.
    Reply
  • yacoub - Friday, December 02, 2005 - link

    Expect few here to understand your point. Many people here just want dick measurements of the largest, fastest pieces of hardware with as few real world constraints as possible. It makes no sense for the consumer but it makes more entertaining reading and they're here for entertainment not research purposes. :[ Reply
  • bob661 - Friday, December 02, 2005 - link

    quote:

    It makes no sense for the consumer
    Makes perfect sense to me and, apparently, a lot of other folks here. I don't think it's out of line at all to test these drives. We ARE enthusiasts here and quite a number of us can afford enthusiast budget systems. If you want J6P reviews go over to PCWorld or something. Granted I'm not in the 7800GTX 512MB crowd (I'm a bang for the buck guy), but I DO like to see these things get reviewed. "I" like to know what's out there.
    Reply
  • Cygni - Friday, December 02, 2005 - link

    Anandtech can only test what the companies send them... which is why you only see reviews of high end parts or parts with special features. Makes things difficult because they could obviously make much more comprehensive and equal comparisons if they got to pick and choose... but thats life. Reply
  • Questar - Friday, December 02, 2005 - link

    No.

    Anandtech *chooses* to only test what companies send them. There is nothing preventing them from ording drives for testing.
    Reply
  • PuravSanghani - Friday, December 02, 2005 - link

    We actually purchased the 500GB 7200.9 the week it was released so that we could benchmark these drives as quickly as possible to let our readers know how it performed before they went ahead and bought it for themselves. We thought that since it was Seagate's highest capacity drive to date and targeted towards servers it would be very informative to those who were looking for high capacity drives.

    This 3-way comparison was written up to show the differences between the highest capacity drives from 3 of the top hard drive manufacturers in the world to help our readers looking for high capacity models to choose the product that best fits their needs.


    Regards,

    Purav
    Reply
  • Visual - Friday, December 02, 2005 - link

    Nah, for products that are already available retail, it was never a problem for AT to buy their own samples. In fact, it's the preferred way, as the companies can't hand-pick better oc-ers or such. This always was one of the pluses of this site. Reply
  • bjacobson - Friday, December 02, 2005 - link

    Anandtech is such a great site. Seems like every other day theres a new benchmark or review of some sort. Far more stuff comes out from AT than many other tech sites. Thanks Anand. Reply
  • johnsonx - Friday, December 02, 2005 - link

    Anandtech does have more new articles per week than most sites, aside from THG. Unlike THG though, Anandtech's articles are quite well thought out, written, and supported. THG's flood of articles every week is more like a bad case of the $hits.
    Reply
  • Visual - Friday, December 02, 2005 - link

    AT, it might be good to include a link to your 7200.9 article ( The so-called mouth-watering failure, http://www.anandtech.com/storage/showdoc.aspx?i=25...">http://www.anandtech.com/storage/showdoc.aspx?i=25... ) and maybe if you've done separate reviews for the other drives link them too in the first page of this one.
    And this reminds me, we haven't yet seen a review of the highest-density 7200.9, the 160GB model. Nor the 133gb/platter 400gb drive :/

    But yeah I know I shouldn't complain, I should just google for reviews by other sites.
    Reply
  • PuravSanghani - Friday, December 02, 2005 - link

    We are currently working on getting our hands on a 160GB 7200.9 unit with the 160GB platters; hopefully within the next month or so.

    Regards,
    Purav
    Reply
  • Zar0n - Friday, December 02, 2005 - link

    Thanx 4 the review, but in the last benchmark, acoustics, u should remove the legend to the bottom so that the bars expanded, and it's easy to see the difference between all drives.

    Waiting for Maxtor DiamondMax 11 to chose the best drive :)
    Reply
  • Slaimus - Friday, December 02, 2005 - link

    It is fairly uncommon for HD manufacturers to release updated firmware after the fact, but that might be what the other drives need. Reply
  • irev210 - Friday, December 02, 2005 - link

    firmware aint gunna help these drives.

    They are all very fast drives... just that the 400gb wd stands out because of its raptor heritage.
    Reply
  • irev210 - Friday, December 02, 2005 - link

    From what I read at storage review, the 400gb western digital has the same chipset/technology as the 10k raptor, just with 100gb plattersx4. That is why you see such high I/O scores.

    I picked up two 400gb RE's (raid edition) to back up my data in a 400x2 raid 1. I couldnt be more pleased with the price/performance. The 400gb drives boast a 1.2 million hours MTBF, which seems pretty impressive for such an inexpensive drive. The motor looks rather large compared to other units, but i was too chicken to take apart my new drives to inspect.

    What really bothers me is that anandtech rounded up the T7k250gb 160gb version which uses different platters than the T7k250 250gb. The 250gb uses 2x125gb platters which greatly improves performance over the 160gb.


    For me, I use the T7k250 250gbx2 in a raid0 SATA 3.0gb/sec with the option of adding 2 more for much better performance and space in a raid0

    for data, I use 400gbx2 raid 1


    the best part is, I have lots of room to grow. I can grow to a 1tb main array and a 800gb data array.

    I hope to see anandtech tackle some more hard drives! The 400gb western digital is an absolute STEAL at the price.
    Reply
  • karioskasra - Friday, December 02, 2005 - link

    How would the WD perform in RAID1 when compared vs the other drives in RAID1? You hinted at it possibly being slower when striped, but how big is the difference? Does it maintain the lead over the other two when striped, or is it outperformed? Reply
  • Cygni - Friday, December 02, 2005 - link

    Makes me chuckle that a 1.5Gbps drive dominated the 3.0's. Not a surprise at all. ATA133 isnt even maxed out by these drives, let alone a SATA150 connection. Kinda reminds me of AGP... or PCI... or pretty much most of the standards these days getting replaced by "better faster must have!" standards that cost the end user money and offer no real improvement in performance.

    Hey, gotta keep selling boards, i guess.
    Reply
  • Griswold - Friday, December 02, 2005 - link

    quote:

    Kinda reminds me of AGP... or PCI... or pretty much most of the standards these days getting replaced by "better faster must have!" standards that cost the end user money and offer no real improvement in performance.


    I can understand mentioning AGP, but PCI? You gotta be kidding me... that bus is such a bottleneck. You dont even have to run PCI cards to find out, just stress all the on-board stuff on a feature rich mobo and you'll notice it too.
    Reply
  • Cygni - Friday, December 02, 2005 - link

    Which explains the rush from the mfts to get PCI-Ex cards out the door. :p Really, the only 2 cards that i can see benifiting from the PCI Express bus are high level RAID cards and gigabit ethernet... both of which are being fully integrated into southbridges anyway. Reply
  • Hikari - Friday, December 02, 2005 - link

    You said AGP and PCI, not AGP and PCIe. Obviously there isn't a lot of difference between the latter, but there is quite a bit of difference between the former. Reply
  • Griswold - Friday, December 02, 2005 - link

    They are integrated into southbridge and still utilize the PCI bus mostly. PCI bus aint only the slot you see on your mobo, you know.. Reply
  • Anton74 - Friday, December 02, 2005 - link

    High level RAID? A single PATA drive has an interface speed identical to that of the PCI bus (133MB/s) these days, all by itself. And then there's SATA with 150MB/s and 300MB/s interface speeds now. Not to mention the PCI bus is usually shared with a multitude of devices, all wanting some bandwidth. Reply
  • puffpio - Friday, December 02, 2005 - link

    It seems anomalous that the Western Digital Raptor 10000RPM drive is sooo much slower in the Doom 3 level load test compared to all the other drives. It sticks out like a sore thumb. It doesn't make sense because it had been dominating the other tests... Reply

Log in

Don't have an account? Sign up now