Hitachi Deskstar 7K1000: Two Terabyte RAID Redux

by Gary Key on 4/23/2007 12:00 PM EST
POST A COMMENT

26 Comments

Back to Article

  • Abki - Friday, May 22, 2009 - link

    RAID 0 only give faster transafer rate of data from/to disk.
    It dosent give anything else faster even if you want it. If time for get or put data to disk dosent matter you have to buy a cheaper disk instead. Why weaste money on raptor or Deskstar.

    About raid 1, it isnt so good and safe. Accidents has happen with 2 disk system, 1 get failure and raid system transfer same fault to the other disk. In the end its no data on any disk that is usuable. The only tru safe is to have a backup of all data. Sould be taken at least every day. That is what the second disc should be used for and not RAD 1.
    Reply
  • Axbattler - Tuesday, April 24, 2007 - link

    I do believe that the main cost of stripping is the risk of failure. For everything though, you get what you pay for: the second drive, and the space associated with having a second drive. It is no harder to backup 2TB in RAID-0 than 2TB of individual drives, if you are going to have that much data, then it is an issue you are going to face anyway. So to me, it comes down to trading off risk of failure, and selected performance gains. I happen to think it is not worth it but I do think it is a matter of opinion (more of a 'to each their own' type of thing). Reply
  • michal1980 - Monday, April 23, 2007 - link

    any one? I know its an extra drive. but you get some data protection, and some performance benefit, and you do not loss 50% of the drive space.

    but I heard bad things about on-board raid 5.
    Reply
  • strikeback03 - Tuesday, April 24, 2007 - link

    I don't believe you lose any drive space in RAID 0, as there is no data redundancy. Reply
  • yyrkoon - Tuesday, April 24, 2007 - link

    The bad things you've heard concerning onboard RAID5 was probably performance related, which would most likely be true, since most onboard RAID would not have a good XOR 'engine', unless perhaps some of the server classed boards have them.

    As for Matrix RAID, I personally am a bit sceptical, from my limited understanding of it, it basically gives you the ability to do a RAID 0+1 array on two drives. Someone correct me if I am wrong.
    Reply
  • tshen83 - Monday, April 23, 2007 - link

    I thought last time I already criticized that reviewing the 1TB drive with RAID 0 performance is missing the point, yet you come up with a redux part on the same drive. First of all, I don't care how much Hitachi is paying YOU or AT to pimp this drive(obviously they paid the wrong guy). If you don't understand that comparing a 1TB drive against the raptors are STUPID, you should get your head checked. One is targetted at the MASS STORAGE people(you know, I should spell it out for you, the guys who want 4TB +) and the raptor is targetted at the people as a PERFORMANCE OS boot drive. COMPARING THOSE TWO DRIVES ARE STUPID!

    now, if you really need a triple redux on this stupid drive, you should cover power usage, noise, heat, and basically constantly read/write to the 1TB drive to see how long the deathstar will last.
    Reply
  • sdsdv10 - Tuesday, April 24, 2007 - link

    The title says it all... Reply
  • Sunrise089 - Monday, April 23, 2007 - link

    Personal Attack: Check

    Totally unsupported claim of bias: Check

    Misunderstanding of market segments: Check

    Use of ALL CAPS: Check

    Improper understanding of "are/is": Check

    Comment that makes you look like an ass: Check
    Reply
  • tshen83 - Wednesday, April 25, 2007 - link

    I will be honest:

    you are right about those: Personal Attach: Check and Use of ALL CAPS: Check

    you are wrong about the following:

    totally unsupported claim of bias? about what? about Hitachi drives being deathstars? Did you read that Hitachi bought the IBM deathstar hard drive business?

    Improper understanding of "are/is"...ok...typo..so what? Try typing the whole message in 20 seconds.....you is stupid :)

    comment that makes me look like an ass? So do you consider all criticisms make people look like ass?

    Misunderstanding of Market Segment: Really? Is it me who misunderstood? The whole point I made is that Anandtech shouldn't review the 1TB drive against the raptors because the difference in market segment. 1TB isn't about performance. It's about mass storage. Mass storage drives needs different measurements: reliability(MTBF), RAID 1/5/6 performance, heat, noise, power consumption. OS performance drives(raptors) need those benchmarks: IO, Seek, Sustained write/read, etc. you get the point. Firmwares are tweaked differently for those two opposing sides.

    I agree with the poster below you, Don't read the review: Check! To be honest, Anandtech hasn't come out with anything good for a while.(except that AppleTV review)
    Reply
  • yyrkoon - Tuesday, April 24, 2007 - link

    LMAO

    You forget one thing on your check list.

    Dont read the article ? CHECK !!!
    Reply
  • sprockkets - Monday, April 23, 2007 - link

    Pic wanted of smoked drive. Reply
  • lplatypus - Monday, April 23, 2007 - link

    RAID-0 is not a general solution to improve disk performance. It solves a specific problem: it increases sequential disk bandwidth. If you're not doing sequential disk accesses, then don't expect RAID-0 to help.

    Of course if your application is doing random disk accesses unnecessarily, then that's a programming problem. But often an application really does need to do random disk accesses.

    Advanced filesystems attempt to make common file access patterns translate into sequential disk accesses, so poor RAID-0 performance might be a symptom of a poor filesystem design in your OS. Yet there is a limit to how much the filesystem can produce sequential disk accesses.
    Reply
  • yyrkoon - Monday, April 23, 2007 - link

    quote:

    However, the test results should give a very good performance indication of RAID 0 without CPU or GPU limitations.


    GPU limited HDDs eh ? Come on guys, this statement is starting to sound as though you think using RAID0 in a gaming system is going to increase FPS or something . . . which, we all know is wrong(or should know).

    Gee, a Raptor smoking its platters, this aught to quiet those people who think the Raptors are impervious to melt downs. Seriously though, how much was this HDD used ? How old was it ? Personally, I think the answers to both the two questions above, would be worth well more than about 20 AT articles . . . and I must say I rather enjoy reading the articles.

    Looking forward to you adding Seagates fastest of their SATA line, and would like to know if you have plans, or would consider testing the NL35 line as well. From what little I have read, the NL35 line are supposed to be geared towards Video, so I would assume the sustained throughput would be higher, but thats why the request for a test of these drives.
    Reply
  • nullpointerus - Monday, April 23, 2007 - link

    quote:

    GPU limited HDDs eh ? Come on guys, this statement is starting to sound as though you think using RAID0 in a gaming system is going to increase FPS or something . . . which, we all know is wrong(or should know).

    As the article itself says, they did so to answer readers' (IMO ridiculous) objections to the previous RAID article. So you should poke fun at the gamer RAID crowd instead.
    Reply
  • yyrkoon - Tuesday, April 24, 2007 - link

    That was not my point, I am fairly confident that AT staff doesnt believe this, but it only leads to confusion, for the less tech savvy. I can hear it all over the wanna be 'geek' 'channels' already . . . 'OMG RAID0 increases FPS in <insert first person title name here>, becasue AT said so . . .'. Even though we all know, this is not what was said. Reply
  • SilthDraeth - Monday, April 23, 2007 - link

    hopefully our second round of results will answer more questions than it raises answers. Reply
  • johnsonx - Tuesday, April 24, 2007 - link

    quote:

    We are back today for a quick look at RAID 0 performance and hopefully our second round of results will answer more questions than it raises answers.


    The last word in that sentence just needs to go away. Then it makes perfect sense: "... answer more questions than it raises."
    Reply
  • joex444 - Monday, April 23, 2007 - link

    I don't find it confusing, but then again I passed grade 4. Reply
  • vailr - Monday, April 23, 2007 - link

    Could you comment on AHCI mode? Comparing enabled in bios vs. disabled?
    The AHCI "hot swap" feature, for example. Seems to work fine when connected to a SATA JMicron controller, yet DO NOT when connected to an Intel port. On the Gigabyte DS3 board, for one example.
    Also: there are newer Intel Inf and Matrix drivers.
    Used for this report:
    System Platform Drivers:
    Intel 8.1.1.1010
    Intel Matrix RAID 6.2.1.1002
    vs.
    Intel Inf Driver Version 8.4.0.1010 Beta
    http://www.station-drivers.com/page/intel%20chipse...">http://www.station-drivers.com/page/intel%20chipse...
    Matrix Storage Manager drivers version 7.0.0.1020 WHQL
    http://www.station-drivers.com/page/intel%20raid.h...">http://www.station-drivers.com/page/intel%20raid.h...
    Reply
  • DigitalFreak - Monday, April 23, 2007 - link

    AT yet again proves the RAID-0 freak'tards wrong. Reply
  • mesyn191 - Tuesday, April 24, 2007 - link

    A POS software RAID controller was used again for the testing though, of course its gonna make RAID 0 look bad, or for that matter RAID 5 too. You need a good hardware (ie. Areca 1210) RAID controller with a CPU and dedicated cache for RAID 0 to be worth while, same goes for RAID 5 or 6. Reply
  • Sunrise089 - Monday, April 23, 2007 - link

    I got my own share of bashing comments in this space for the previous article, apparently price/performance questions aren't as valid as dogma. Anyways, whatever the perception of the community to this article, I think it speaks well of AT that this story ever appeared. Many sites would just let their old article speak for itself, and leave the questions it raised unanswered. You went out and made a new rig and have hopefully answered some of the questions the folks on the fense regarding Raid O may have had. Keek up the good work. Reply
  • poohbear - Monday, April 23, 2007 - link

    very true, that Anandtech follows up on their articles speaks quite highly of this site. cheers and thanks for clarifying 110% what raid 0 should and shouldnt be used for.:) Reply
  • Lifted - Monday, April 23, 2007 - link

    I don't understand why AT mixes these review together. You always end up with people complaining about the compromises being made, and to a certain extene they're complaints are legitimate.

    Make one review for a hard drive, and a seperate review or article on RAID configurations. There are so many possibilities when it comes to RAID configurations that these short reviews can only raise more questions than they answer. You'll always have people saying something about the system, the array adapter, stripe sizes, even the damn GPU. When it comes down to it, people that use RAID where the performance counts (servers) are just going to by an HP, IBM, Dell or whatever system and use the adapter that comes with it. Home users and their 2 or 4 SATA RAID arrays are never going to see or need the performance from these systems that they seem to always be complaining about in their responses to these reviews. Is there a reason they need 180MB/s rather than 140MB/s to store mp3's and movies?
    Reply
  • Eastbay2359 - Monday, April 23, 2007 - link

    "The drive literally smoked its platters. Of course we lost the entire test image and a significant amount of test time"
    WHAT NO BACKUP !! :)
    oh, from the previous paragraph
    "a data backup nightmare"
    Reply
  • yacoub - Monday, April 23, 2007 - link

    yowza! so compared to smoking the tires in a sports car, apparently smoking the platters in a harddrive is NOT cool. ;) Reply

Log in

Don't have an account? Sign up now