raid specific drives

[protias]

i have read how WD advertises raid specific drives for their 400 and 500gb drives. i am wondering why these drives are not really raid type drives. what makes up a raid specified drive?

*hopefully no one tells me to search, because i just don't have that kind of time on my hands* :(

edit: these drives would of course have a raid 5 controller

[unhappy_mage]

The key phrase is TLER: Time Limited Error Recovery. The problem with other WDs is that when they encounter an error and must recover from it, they stop responding to commands from the controller. With raid controllers, this can mean that the controller thinks the drive is unplugged and drops the disk from the array. Thus, when the disk starts responding again, it's got to be re-added to the array - controllers can't trust what's on a disk to stay on it when you take it out of the array. This causes a lot of reading and writing, and if another disk happens to drop during this time the whole array is toast.

Desktop drives are still better for the desktop - there, you want to have the error recovery take place, since there's no other source for the data to come from.

http://www.hardfolding.com/ftag1.php/mem/150072.png (http://www.hardfolding.com?go=38&tm=33&id=150072)http://www.hardfolding.com/utag1.php/mem/428/1.png (http://www.hardfolding.com?go=36&id=428&type=1)

[protias]

so it would be better to go with regular drives instead of the TLER drives if i wanted to do a raid-5 of 10+ drives? considering from the standpoint of having the controller.

[drizzt81]

no, you'd want the RE-drives in that case. You'd want the regular drives if you were doing a 4 disk R-0 array or using them in a desktop. Then again, it also depends what controller you are going to be using for this.

[protias]

this controller (http://www.newegg.com/product/product.asp?item=N82E16816115024) with drive stuffed in here (http://www.newegg.com/product/product.asp?item=N82E16817121405) and that stuffed into this case (http://www.avadirect.com/product_details_parts.asp?PRID=6143). i have 3 of the 4 into 3 (http://www.case-mod.com/store/cooler-master-stb3t4e1-stacker-module-p-1523.html), but when i saw the 5 into 3 (http://www.newegg.com/product/product.asp?item=N82E16817121405), i knew i had to have 3 of those instead. i am waiting for prices to drop (and to save up LOTS of money) for probably 18 400 (http://www.newegg.com/product/product.asp?item=N82E16822144424) / 500 (http://www.newegg.com/product/product.asp?item=N82E16822136032) gb drives. and that includes a spare drive.

[TeeJayHoward]

this controller (http://www.newegg.com/product/product.asp?item=N82E16816115024) with drive stuffed in here (http://www.newegg.com/product/product.asp?item=N82E16817121405) and that stuffed into this case (http://www.avadirect.com/product_details_parts.asp?PRID=6143). i have 3 of the 4 into 3 (http://www.case-mod.com/store/cooler-master-stb3t4e1-stacker-module-p-1523.html), but when i saw the 5 into 3 (http://www.newegg.com/product/product.asp?item=N82E16817121405), i knew i had to have 3 of those instead. i am waiting for prices to drop (and to save up LOTS of money) for probably 18 400 (http://www.newegg.com/product/product.asp?item=N82E16822144424) / 500 (http://www.newegg.com/product/product.asp?item=N82E16822136032) gb drives. and that includes a spare drive.
A note - You won't get the failed disk leds on that enclosure to work with this card. Or, you won't get the drive activity leds to work. You'll have to choose which leds are more important to you. (I have that enclosure, and a 2220 card.)

[drizzt81]

sidenote: ZZF often times has better HDD prices than newegg. Whenever you decide to buy, make sure to check them out too.

[protias]

sidenote: ZZF often times has better HDD prices than newegg. Whenever you decide to buy, make sure to check them out too.

i had a very bad experience with ZZF recently. their return policy for defective merchandise is vague and wouldnt give me a full refund ($7.xx) of the remaining purchase. after 2-3 months of failing to get through their tiny brains, i complained on reseller ratings and was refuned a few days later. they will never have me as a customer ever again.

[protias]

A note - You won't get the failed disk leds on that enclosure to work with this card. Or, you won't get the drive activity leds to work. You'll have to choose which leds are more important to you. (I have that enclosure, and a 2220 card.)

personally, i'll take the LEDs to show me which ones have failed. im perfectly fine with having no LEDs showing when working properly. thanks for the heads up

[drizzt81]

i had a very bad experience with ZZF recently. their return policy for defective merchandise is vague and wouldnt give me a full refund ($7.xx) of the remaining purchase. after 2-3 months of failing to get through their tiny brains, i complained on reseller ratings and was refuned a few days later. they will never have me as a customer ever again.
that's fair enough. I figured that I should mention it, since I have had good experiences with them. But I completely understand you reasoning. In the end, I would likely act the same way.

Good luck with your build. I am already jealous ;)

[protias]

that's fair enough. I figured that I should mention it, since I have had good experiences with them. But I completely understand you reasoning. In the end, I would likely act the same way.

Good luck with your build. I am already jealous ;)

lol, im jealous too. the only problem is that im short by about 3k to do this project. so as u can tell, its not happening for a while :(

[StorageJoe]

I'd actually recomend the Seagate NL35 drives. They are the only "enterprise sata" drives that offer RV attenuation. The WDs may be fast in a one or two drive system, but put 12 of them in a box and they just tank, while the Seagate's shine. We've tested several drives and now use nothing but Seagate in multi-drive installs for just this reason.

[TeeJayHoward]

I'd actually recomend the Seagate NL35 drives. They are the only "enterprise sata" drives that offer RV attenuation. The WDs may be fast in a one or two drive system, but put 12 of them in a box and they just tank, while the Seagate's shine. We've tested several drives and now use nothing but Seagate in multi-drive installs for just this reason.
Any chance you could enlighten me as to what RV attenuation is, and how it affects multidrive arrays?

[protias]

I'd actually recomend the Seagate NL35 drives. They are the only "enterprise sata" drives that offer RV attenuation. The WDs may be fast in a one or two drive system, but put 12 of them in a box and they just tank, while the Seagate's shine. We've tested several drives and now use nothing but Seagate in multi-drive installs for just this reason.

this gets my curiousity going (bad because it may kill me like it did the cat).

can you explain to me how a seagate is different in a raid array compared to a WD? wouldnt both use the same amount of bandwidth on the controller?

thanks for opening a can of worms :p

[TeeJayHoward]

Storage Review shows the drives being a little bit slower in Random Access Writes, and faster in Reads, than a WD drive. (http://www.storagereview.com/articles/200510/ST3400832NS_2.html) Faster STR, though. It's inferior in single and multiple user tests, superior in noise-generation and power requirements compared to the WD. Judging from the review, WD drives are superior. I'd REALLY appreciate a link on RV attenuation. Hmm. Matter of fact, I think I'll google it now!

Edit: Rotational Vibration (RV) attenuation - Something to do with putting a bunch of drives right next to each other and not having the vibrations emitted by the spinning platter affect the other drives. Linky (http://www.earsc.com/pdfs/engineering/literature/Engineernotebook/EN502rvtesting.pdf)

[drizzt81]

isn't RV the equivalent to WD's RAFF:

http://www.wdc.com/en/products/products.asp?driveid=158&language=en

[TeeJayHoward]

isn't RV the equivalent to WD's RAFF:

http://www.wdc.com/en/products/products.asp?driveid=158&language=en
Certainly looks very similar. StorageJoe, any chance we could look at the comparison you made between various drive manufacturers?

[Vertigo Acid]

Indeed, I was under the impression that the RE drives, along with the Raptors, had additional provisions for vibration concerns.

[StorageJoe]

When we were qualifing these drives, one of our engineers did some pretty extensive testing of both drives. In a single drive test, the WD was indeed superior, but as soon as you throw 12 of them in a chassis, the Seagates were significantly faster. Mind you this is only the NL drives, as the standard (AS) desktop drives performed no better than the WDs in this environment. From my understanding, the NL drives detect and compensate for vibration in the drive. They do not reduce vibration, but are better able to compensate for it and keep the head on track. I had some pretty slick graphs of the performance data he collected, I'll try to dig them back up.

[protias]

When we were qualifing these drives, one of our engineers did some pretty extensive testing of both drives. In a single drive test, the WD was indeed superior, but as soon as you throw 12 of them in a chassis, the Seagates were significantly faster. Mind you this is only the NL drives, as the standard (AS) desktop drives performed no better than the WDs in this environment. From my understanding, the NL drives detect and compensate for vibration in the drive. They do not reduce vibration, but are better able to compensate for it and keep the head on track. I had some pretty slick graphs of the performance data he collected, I'll try to dig them back up.

that would be pretty cool to see.

[StorageJoe]

Found one of them, how can I post a picture?

Also, after talking to our engineer, it's not the NL series drive that was tested. We got a few eval drives for their new ES line (Enterprise SATA) that were tested vs the WD and Hitachi.

[drizzt81]

Found one of them, how can I post a picture?

Also, after talking to our engineer, it's not the NL series drive that was tested. We got a few eval drives for their new ES line (Enterprise SATA) that were tested vs the WD and Hitachi.
you need to host it somewhere and then post a link with teh [ img ] tags

[unhappy_mage]

If you can't find hosting, email it to me (hardfolding@users.sf.net) and I'll host.

http://www.hardfolding.com/ftag1.php/mem/150072.png (http://www.hardfolding.com?go=38&tm=33&id=150072)http://www.hardfolding.com/utag1.php/mem/428/1.png (http://www.hardfolding.com?go=36&id=428&type=1)

[StorageJoe]

Sent...

I forgot to mention, the part numbers on the ES drives are ST3750640NS and ST3500630NS

[unhappy_mage]

Here is a series of graphs measuring recorded RAW I/Os at various RAD levels.

This is the ES SATA line that they are going to be shipping soon. The main differences between the current NL drives and the ES drives we tested are an increase in areal density from 111 GB / in^2 to 130GB, which jumps the platter size from 160 GB to 187.5GB. Seek time seems to be identical, although transfer rates are improved from about 70 mb/s sustained on the 500 GB NL to 72 mb/s sustained on the 500GB ES and 78 mb/s on the 750 ES. The single drives were producing about

12.5 RAD/sec^2 on the new ES, I don't have measurements for the other drives.

-StorageJoe
http://will.incorrige.us/pics/SR.pnghttp://will.incorrige.us/pics/SW.png
http://will.incorrige.us/pics/RR.pnghttp://will.incorrige.us/pics/RW.png

I just PNGified the files and posted them. Sorry for the delay.

http://www.hardfolding.com/ftag1.php/mem/150072.png (http://www.hardfolding.com?go=38&tm=33&id=150072)http://www.hardfolding.com/utag1.php/mem/428/1.png (http://www.hardfolding.com?go=36&id=428&type=1)

[protias]

wow, thats just amazing on how slow the WD 500gb drives get

[unhappy_mage]

Agreed, it's pretty one-sided. How were those drives secured? How much total displacement was seen? 12.5 rad/sec^2 means that if you were to put that much acceleration in one direction, you'd get the drive turned around 2/3rds times, more or less. On the other hand, if the rotational motion reverses itself twice in 1/120th of a second (as I'd expect with a 7200 RPM drive - whatever causes the unbalance moves from start to halfway across the platter, and back, 7200 times a minute) the distance traveled could only be:
x(t) (displacement from origin at time t, in rad) = A sin(120 pi t)
d/dt x(t) = 120 pi A cos(120 pi t)
d^2/dt^2 x(t) = -(120 pi) ^ 2 A cos(120 pi t)
maximum values of d^2/dt^2 x(t) occur at t=0, pi, 2 pi, etc:
-(120 pi)^2 A * 1
Given the maximum value of d^2/dt^2 x(t) = 12.5 rad/sec^2 the max displacement A is about 87 microrad, not a whole heck of a long way. Is my math wrong (probably :p) or what's going on here?

My theory is that testing the drives on a standalone platform like that means that they can vibrate more than they would in an enclosure, simply because there's less mass to vibrate. Testing how much vibration was actually generated by a number of driveswould be useful, as well.

Also of note: Maxtor Atlas 10k V drives are specced for 7,000 rad/sec^2 of rotational shock, and the other figure I found was 15000 rad/sec^2 (nonoperating) for an older 5400RPM Maxtor unit. Are your units in krad? (lol, k-rad)

http://www.hardfolding.com/ftag1.php/mem/150072.png (http://www.hardfolding.com?go=38&tm=33&id=150072)http://www.hardfolding.com/utag1.php/mem/428/1.png (http://www.hardfolding.com?go=36&id=428&type=1)