$$$$... WD Enterprise HD - 2TB RE4 vs 2TB RE4-GP (x4) for RAID 6

Discussion in 'Mac Pro' started by UltraNEO*, Oct 13, 2009.

  1. UltraNEO* macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
    #1
    Has anyone here either used or using those WD RE4 2TB (x4) or WD RE4-GP 2TB (x4) drives as part of a RAID6 Array? Oppose to my current RAID5 configuration that consists of WD 4 RE3 1Tb (x4) (total 2.79TB). I'm planning to migrate to a more secure RAID level 6 with new drives... So, how are those BIG 2TB RE4's? and which is faster?

    I'm desperately in need of more data space, my RE3 array is down to it's last 84Gb. Yeah I know it sound like alot but man, since my last project the data demands have grown 340Gb and it's still incomplete. Sooo... I wanna upgrade before I DO run outta space! :eek::eek:
     
  2. UltraNEO* thread starter macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
  3. J the Ninja macrumors 68000

    Joined:
    Jul 14, 2008
    #3
    I think the GPs are Caviar Green's with firmware tweaks for RAID array use. Basically, same differences as Caviar Black vs normal RE3/4. I couldn't find anything about this on WDs site, but you might want to check the warranty on those. Caviar Greens are 3 years, whereas Caviar Blacks and regular REs are 5 years. Not sure what these are.
     
  4. UltraNEO* thread starter macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
    #4
    Their RE3 and RE4 are the normal enterprise drives, created for demanding servers/workstation uses as oppse to the home desktop. Timings have been fine tuned for arrays, making them more reliable in the long run hence the price hike but the new ER4-GP is their new, more environmentally friendly enterprise version. However, what I wanna know is their performance differences, if any. If they're slower, how much slower??? I can't seem to find much details about the two running side by side, that's why I thought I'd ask people here...
     
  5. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #5
    I haven't had the opportunity to use them yet. Heck I haven't even seen reviews with performance data on the 7200 RE4 2TB's yet (I need to look again).

    That said, you will notice a difference. The RE4-GP is a Green Power, so it's essentially a 5400 rpm disk. The outer tracks will definitely be slower than the 7200 rpm models. 5400 is 75% slower than 7200 in terms of rpm, and the performance should follow fairly closely, as they should be the same otherwise (same heads, platters, and electronics).
     
  6. UltraNEO* thread starter macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
    #6
    I find this forum weird.

    Every damn week new threads appear askin' for assistance in regards to setting up an array; yet, here I post a thread regarding enterprise drives and there's appears to be zero interest. Ain't you curious Nano? Ain't you curious about what sort of drive those folks are using in their arrays? Did i miss something? or does everyone other brands (Seagate stick-a-brick Hitachi) to WD? Maybe someone here has gone with Intel and built a fast yet teeny tiny array?

    /end rant


    Umm.. Where does it say those RE4-GP's are 5400's? I think I'm going blind cause there's no mention of 5400rpm on WD site... ><
     
  7. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #7
    Most are running their arrays off the logic board, and use consumer drives it seems. Makes more sense here though. You don't have the recovery or stability issues you do with a RAID card. Not all drives are equal, but in the case of WD's, at least their consumer models are rated at a UBE of 1E15. It's just missing the additional sensors and TLER settings (which aren't needed in software RAID).

    It doesn't explicitly say what rpm the drive operates on, and their literature lends one to think it's variable. That doesn't seem to actually be the case though. To save on power, they have to lower the rpm, but 5400 would make sense, and the performance of the past versions of the Green Power versions lines up with that spindle figure (it couldbe spinning a bit higher, perhaps like the Spinpoint Green's at ~5900 IIRC, according to Tesselator). But there's no indication it's variable, or hits 7200 at any point. :(

    If you search around, there's an article or two on it I think. Something about WD having to change some information on the specifications (can't remember the details on the exact line removed, or have a link :eek:). The main part that stuck with me, is the fact they advertised in a manner that makes buyers think it's variable, when it really isn't.
     
  8. goodcow macrumors 6502a

    Joined:
    Aug 4, 2007
    #8
    At home I just use a Drobo and a DroboPro.

    4x1TB WD Greens in the Drobo
    8x1TB WD Greens in the DroboPro (with dual disk redundancy)

    Then at work we have a 24TB Fujitsu Eternus SAN in RAID6 with two-hot swap spares and another DroboPro with 8x2TB WD Greens.

    I love my Drobos. I debated getting the "RE" line of drives, but the cost difference over 8 drives for my Pro at home were significant, and with dual disk redundancy and the critical stuff backed up again to the standard Drobo, I didn't need it.

    I would've bought the RE4 line at work though, but the 2TB models weren't out at the time.
     
  9. UltraNEO* thread starter macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
    #9
    Well, a long time ago I tried the consumer route with Seagate Baraccuda 7200.11 then WDC Blacks, created an array on the cheap, unfortunately it didn't workout for me:mad: I learnt the hard way! It's not profitable to loose almost 2Tb's of edited media with the deadline closing in fast! I lost hours of leisure-time all due to a little crash, was forced to recreating everything from a two week old backup, getting stressed and losing sleep in the process:(:(:(

    Well there have been loads of benchmarks for the RE4-GP vs other drives (e.g: 1, 2, 3), same can be said for the RE4 but non (i've seen) compare the two side by side... Odd... Are they aimed at different ends of the professional market???
     
  10. UltraNEO* thread starter macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
    #10
    Those Drobo's are great little devices, unfortunately for an online scratch disk they're a bit slow as it's data transfer rate is very much limited by it's initial connection (USB2, FW800, 1Gbit LAN).

    Where as with an internal RAID connection via a MiniSAS (iPass); i'm technically connecting each drive directly to the interface card via a single connector, increasing the bandwidth; it's as though each drive has a eSATA connection. The rest is ofcourse limited to the interface card's onboard processor... and Nano will tell you all about that given the opportunity.

    Mmmmmm.... I thought about investing in a SAN rack but it's serious money! And it's serious money populating the damn thing after! I don't have that kind of cash right now. Spent it on a nice motorcycle ('09 ZX-6R9PF)

    [​IMG]
     
  11. goodcow macrumors 6502a

    Joined:
    Aug 4, 2007
    #11
    I forgot that I also have a software RAID 0 with my MacPro at home using two, 320GB WD RE3 drives. I bought them to capture 720p and 1080i uncompressed, but actually I find the speed to be insufficient at times. I need to throw a third drive in there.

    Yeah, our SAN was about $25,000 and that's likely with a discount as I work at a government agency. The drives alone are about $700 a pop for Fujitsu nearline-SAS 1TB models.
     
  12. thederby macrumors regular

    Joined:
    Jun 22, 2007
    Location:
    Austin, TX
    #12
    please don't use disk (even if it's redundant) for critical backups...
     
  13. goodcow macrumors 6502a

    Joined:
    Aug 4, 2007
    #13
    Okay, by "critical" I really mean important stuff, as I'm a home user who loves media and not a business.
     
  14. UltraNEO* thread starter macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
    #14
    Sound advice.
    Me, I've turned to Blu-Ray.

    ... so far i've only managed to create one expensive coaster! LOL
     
  15. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #15
    I've never been that much of a fan of enterprise drives, as you've seen in other posts. It's hard to get users to understand the differences, and most end up getting the consumer models, as they're less expensive, and easier to find. You can have problems with them off the logic board (system board controller, no matter who made it). But when they're on the logic board and not a RAID controller, they tend to have much fewer problems. But it's not always non-existant (but usually along the lines of single drive issues; DOA's, poor firmware, or bad cables/loose connections).

    The only thing I'm aware of you using, is the CalDigit + RE3's. BTW, have you traded it out the card yet?

    I know there's plenty of benchmarks on the RE4-GP models, but I've not found what I've been looking for yet on the WD2003FYYS (7200rpm 2TB model). It all still seems to be product announcements. :rolleyes: But when it does come out, then you'd have to take performance data from each (almost certainly from different sources), and compare the results. But I can tell you this: The GP models will be slower. :eek: :p

    BTW, I found this:
    (Source). This link has some super basic data on the WD2003FYYS, but rather limited. Their conclusion is it's great at large files, but not so wonderful on small ones. But not really any throughput data.

    As it happens, their conclusion does make sense, if the Barracuda is running faster (5900 vs. 5400). They just don't provide much detail, to let a reader know how they stack up against one another over various tests. And I don't trust Seagate that much yet either. So personally, I might opt for the WD on the expectation of fewer problems given their recent track record vs. Seagate. But we still need more information. I can't even find a source in the US either, not just a lack of throrough benchmarks. The only thing that really pops up (non WD sites), all seem to be based in Australia.
     
  16. DoFoT9 macrumors P6

    DoFoT9

    Joined:
    Jun 11, 2007
    Location:
    Singapore
    #16
    i think the fact that its "IntelliPower" indicates that its 5400rpm normally but can spin up to 7200rpm when needed. From this article:

    ultra: after reading this article i am not entirely sure if i would go with the re4-gp drives. have a read and see for yourself.

    the fourth post down in the above article by "John" seems to contradict some of the information given to us, hmm.
     
  17. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #17
    That's off of manufacturer's data though. I've not seen any independent confirmation that that's actually the case. It would make more sense that they tested out the motors (rpm) to reach a power goal, and set it at that speed. Seriously.



    I'm with "John" on this one. Firmware is my first suspicion with the issues, followed by settings used. Either that of the drives or the card, and that's what the various card makers and WD are working on.

    Unfortunately, the "We aren't experiencing problems" response is becoming too common. It reminds me of the same thing the OP in that thread recalled with Seagate and the ES.2's, not just the consumer models. I'm hesitant to trust Seagate easily ATM.
     
  18. DoFoT9 macrumors P6

    DoFoT9

    Joined:
    Jun 11, 2007
    Location:
    Singapore
    #18
    You think its just a marketing compaign? And the actual results will be different? That might be so but if somebody external tested the speeds then they would get in some serious trouble. So on this one im going to stand by my statement until proven wrong.

    It would make sense. but i cant imagine why its so hard to continue supporting hardware that they are already supporting..

    Oh i have had a lovely experience with apple lately and the 'we arent aware of this problem' in relation to a time capsule, so i know the feeling.

    Seagate has gone down hill, both in reliability and customer interations. They only have themselves to blame i guess.
     
  19. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #19
    Yes, I do. Think about it. There's more work involved to figure out the curves (power/rpm), and then create the algorithm in the firmware. Yet the drives are less expensive than their 7200 fixed rpm counterparts.

    To independently test it, you'd have to actually measure the platter rpms to be truly accurate. There's other ways you could figure it out perhaps, but you'd need a drive as a sacrifical lamb (disassemble the motor, and measure the voltage requirements for a given rpm) to derive it from the motor input voltage. Anything else, and you could end up with an inaccurate test (i.e. vibration sensors to prevent the disassembly/possible destruction of a drive).

    Issues come up though. I've seen drives placed on the Compatibility List, the card maker gets a bunch of reports for something they didn't experience during testing, then pull it back off the list. Eventually they get it sorted, and it goes back on, usually with a new firmware revision, either from the drive maker or for the card.

    Then there's the problem that card makers can't test every drive out there. Just too many, and they can only test what they have available to them. Then take age into acount. :eek: Tested vs. untested drives end up in quite a disparagement due to practical realities.


    It's getting much more common I'm affraid. :( Budget cuts and the business end is in too much of a hurry to obtain funds in hand. Simply put, they want their money back ASAP, and assume the product will work after the preliminary work is done, but not complete. They assume the engineering and support staff can resolve any issues after it ships. They don't seem to understand (or more likely care), about the damage done to their company's reputation by such antics. My advice, is vote with your wallet in such cases, and get the word out, so others won't buy. I'm picky about such things, and don't want others to go through the headaches if I can help them avoid it.

    Yes, but they are the only ones, nor will they be the last. :eek: :p
     
  20. DoFoT9 macrumors P6

    DoFoT9

    Joined:
    Jun 11, 2007
    Location:
    Singapore
    #20
    fair call, if i were true the prices probably would be higher as more research/testing would have been put into the drives. i dont see why they would lie though, but seeing as though they dont have a clear definition on their website then i guess we cant call it lying :rolleyes: (i couldnt find a page with a definition, am i missing something?)

    sounds complicated. im not volunteering to do it :p

    for new hard drives (such as the RE4), would they be somewhat similar in operating to the older (RE3s?) hard drives? are the changes THAT large that it needs a complete upgrade?

    oh yes of course, i guess they would generalise a fair bit. e.g. "the card works with this drive, so it will have to work with its v1.1" or something of that sort.



    oh and dont i know it. the issue i have (and EVERY other Time Capsule owner) is that the capacitors in the power supply are exposed to too much heat. apple tried to make a small unit with no ext. PSU, which they did beautifully - but they didnt give it any airflow to keep it cool! so yes. blown capacitors after 18months.

    for now, i am advising that NOBODY buys a TC until they fix it. there is no use really. its going to die eventually.


    do you mean they arent the only ones? that would make more sense :p

    get some sleep eheh.

    is that test conclusive? (the article i linked to earlier). the write speeds are pretty darn slow really... reads are ok.
     
  21. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #21
    I've not seen a definition either, and suspect it's intentional. It allows them to be vague, not only to the consumer, but keeps their IP out of the hands of their competition as well.

    At best, if it actually occurs, the differential in rpm is very likely lower (i.e. it floats between say 5900 and 5400), as the spindle is the power hog in a mechanical drive, followed by the servo, and the electronics are a distant 3rd (which is why SSD uses so little power in comparison).

    Not so much. Newer designs are usually based off the previous ones, and in the case of the WD2003FYYS, it's likely due to the lateness of it's release that it gained the RE4 designation rather than the RE3, as it seems the difference is only the platter density has changed (what the RE4 line offers btw, so it fits perfectly). It would appear as "old tech" had they given it the RE3 label. ;)

    Take a look at a few. You'll see a card model #, a drive model #, and firmware revisions for each. So detailed research is required before purchase. Where it gets hard, is if you're not getting a complete system (changing either the card or drives alone).

    Somehow, I'm not really surprised. If I had to guess, they didn't use the right capcitors though in terms of temp ratings. 105C can be had, but they likely used lower. 80C at best, and keep in mind, as the temp rises, the cacitance is reduced, further increasing the stress on the device. The rated value is usually that at ambient (25C). To know what it will do in a design, you need to heat it up, and test a few parts to create a curve. Not all manufactured component makers do this for you. :rolleyes: A hint as to how well it will perform too. :p

    Not a sleep issue, just typing too fast. :eek:

    Not really, as it's one case (and isn't working as it should). If there were more experiencing the same issue, then yes. Other tests, on similar systems that are functioning properly, would be a better indicator. Checking other sources, such as Storage Review would hopefully offer additional information.

    However, given the issues are with multiple cards, I'd wait before using them. At least do as much research as possible, to see what's going on. Assuming it isn't an isolated issue (which would seem to be with the system in question, and it does happen), waiting for a solution to arrive would be the safest course of action. If the need is immediate, then look to the competition, and you have to do the research for each drive and card combination you're intersted in. Research, research, research. RAID's full of it. :p
     
  22. DoFoT9 macrumors P6

    DoFoT9

    Joined:
    Jun 11, 2007
    Location:
    Singapore
    #22
    seriously, what are the benefits? are the power increases between these two THAT great? (questions arent aimed at anybody ha).


    there shouldnt really be problems then should there? as there are no changes with how it works/reads/writes/stores/etc


    research shows that it idles at about 60°C inside that little container, the fans are not programmed properly and rarely turn on. the bottom is rubber and doesnt let heat out, there are tiny tiny air holes. the PSU is apparently a very good one (forget the brand) the capacitors are apparently high quality too (also i forget the brand lol). but yea..

    i am going to replace the blown capacitors. cheap and simple.


    correct your typing skills then ;)


    the reason i thought it would be a possible problem is that hes doing so many tests with quite a number of the HDDs, if it were one or two then yea that would be suspicious.

    the guy did say he just pulled them out of the box and didnt change them at all, maybe there was a firmware upgrade? seems likely.

    p.s. why am i now Dr. Pants in your last reply nano lol?
    p.s.s. how can one discharge a capacitor safely :p
     
  23. nanofrog macrumors G4

    Joined:
    May 6, 2008
    #23
    In terms of a single drive, it won't seem like that much. But the drives are aimed at the enterprise market. Think many multiples of drives in racks in a data center. Take that small amount, mulitply by the drive quantity, and figure out the power savings. Now add in the lower cost of the HVAC's power usage, as it doesn't have to run as hard to cool them. Search around, and you'll notice it, in such terms as IOP/Watt for systems, ... It adds up fast, and businesses really do watch that, as the power bill is in the thousands of USD per month. :eek: :p

    Not so much. But the GP's do use different firmware than the performance counterparts, and firmware is easy to botch (firmware of the RE4-GP /= to that of the rest of the RE4 line).

    RE2 to RE3 was more of an issue, as the drive electronics did change from a single processor to a dual processor controller chip. This resulted in the slower addition of the RE3's to the lists IIRC. More firmware revisions to make it work.

    I didn't realize it was that bad for airflow. But if they were designed properly, it shouldn't be failing at 60C, using 80C parts (assuming the capacitance was taken into acount at the working temp). My guess is it wasn't at all.

    I'd recommend using the 105C versions (80C as a minimum, but you may not be able to fit what you need), and test the part first to get the capacitance close. Even going back and determining the correct value for the design yourself, as it's got to be lower than it should (i.e. rounded down, then didn't consider the actual value at working temp). It makes a difference.

    Keep in mind, you're using a fixed board (device used) and internal space, so pay attention to the dimensions of both (namely the capacitors used), as I've no idea if you can go with a larger diameter or height.

    Why? :eek: I made you use your brain to fill in the correct term given the context. :D :p :p



    Understandable. That's why it needs further investigation.

    And you're on the money with the second part. It appears he never checked the drive's firmware revision, or the cards. Let alone ever tested the drives individually (i.e. SMART, bad sectors,... it even gives the firmware revision) before implementing them in an array.

    It allows you to determine the health of the drives before you start, and update the firmware if needed (more common when you get a drive shortly after release btw). This adds time to the initial implemtation, but it's really not an option. Then figure in all the testing needed before trusting your actual data to it. :rolleyes: ;) It takes time. Lots of it, especially if everything's new.

    Oops. I must have copied "
     
  24. DoFoT9 macrumors P6

    DoFoT9

    Joined:
    Jun 11, 2007
    Location:
    Singapore
    #24
    true i never thought about it on such a large scale (silly me). save 1-2w per drive, if you have 1000 of then that is a massive savings.


    the RE4-GP are the "energy saving" ones and the RE4 are more the "performance" ones. got it.

    ahh i see. i thought that would have been up to the HDD manufacturer to make them work properly. clearly not - they just make the thing then leave the others to make it work.


    tbh the airflow is non-existent. the rubber (which i assume is to stop vibration) is glued onto some metal that has some really good airholes (0.5cm in diameter) every 1cm or so. but the rubber glued to it doesnt do anything of course. to fix this i am simply not going to stick the rubber back on, the results will be good i hope.

    a singluar capacitor is so hard to find around here let alone the higher end ones. ebay wont work in this case as i want quality.


     
  25. UltraNEO* thread starter macrumors 601

    UltraNEO*

    Joined:
    Jun 16, 2007
    Location:
    近畿日本
    #25
    Despite your disliking for them, I already have a buyer for the CalDigit card.. Yay!! And the new replacement card is on order!!! Seems it's such a specialist item. It took a little while to find a local dealer with them in stock! LOL.

    I took your advice, Nano, did some research and gone for something that has more expansion than I need, this way I'm covered and I'll have room to expand. Though I do think this controller card will be out of the budget for most users (Areca ARC-1680ix-24 PCIe + 2Gb Cache upgrade). Been told by so many network administrators that this thing isn't just fast, but blazingly fast!!

    [​IMG]

    I think the next task is to figure out how I can add more drives to a already 'full' Mac Pro.. Time to strip that PowerMac G5 2Ghz :D:D and make it actually useful, while it's occupying my living space :) and buy some quality hard drives!! I'm thinking six for now is plenty.
     

Share This Page