Bought the Qnap TS-451 - Which Drives?

[Jaybo]

OK, Found the HGST 3TB NAS drives for $124 ... Thinking I should probably jump on that price ... Of course if I do the dang things will probably go on sale

 

[Jaybo]

For a more current picture of the reliability rates as well:

https://www.backblaze.com/blog/hard-drive-reliability-stats-for-q2-2015/

 

[sfx2000]

Now that you mention it, that might be another argument in favor of using a 5200 or 5400 RPM HDD in a NAS rather than 7200 (beside the obvious heat/power aspect).

Would be interesting comparing the typical life expectancy of a WD Red vs WD Red Pro (5x00 vs 7200 rpm).

Spindle speed - not as big of a deal as folks might think - most of my 2.5" drives are HGST 15K SAS drives, running Oracle... in an environment that is fairly busy (think 10M users) - they run 24/7, and I see one or two fail a year... they're not high capacity, but trust me, a 15,000RPM hard disk, esp. when you have a bunch of them - it's like standing next to a jet plane...

That's getting off track a bit...

With the newer very high density drives out there in SATA 3.5 in space, there isn't really a need in our market segment, the SOHO space, for much more that 5400RPM or so... the number of bits flying under the head in a sustained write or read is impressive - 7200 can help with some applications where random read/write performance is a higher priority (like Database work for example), but generally, most folks might not benefit in a NAS box with anything higher speed, it's more important to look at the SATA levels and platter sizes..

The backblaze stats were interesting on the number of failures across vendors - should also note that most of their drives were off-the-shelf USB drives that they shucked the housings on - they try to justify this to some degree, by suggesting the enterprise class drives costs more (they do) and don't fail as often (which they also do), but considering cost vs. MBTF and real-world failure rates, they can tolerate the risk due to their filesystem implementation is...

And - they have a specific application profile, and they're trying to do it as check as possible - crowdsource USB external drives, ship them under unknown circumstances, and then remove the USB housings and drop the bare drives into a pod - and they don't disclose the average usage and operating environment (temp/duty cycle/etc) - they just see one vendor having more problems than another...

Not very scientific... kind of reminds me of the recent Samsung SSD scare with TRIM, where the blog post absolutely said that SAM has a problem with Trim functionality, and it ended up being a Linux Kernel issue with split block IO tasks...

So all - please consider this when posting/reposting the backblaze link... they haven't done all their homework, and more importantly, they haven't even suggested that they've reached out to Seagate (or other vendors) to analyze the failures.

Ever vendor gets a bad run of drives - even HGST, which everyone here seems to be a fan of (they're a good vendor, I use them), but HGST is the child/origin of the IBM 75GXP, which at the time, the Deskstar was the best performing drive of it's class, but failed well beyond any expectations - e.g. this is the Deathstar drives.

And yes, WD has had their fair share of problems as well...

 

[stevech]

what about these NAS-specific drives where the rotation speed varies (I suppose by track location)? Seems WD red et al are taking this to trade secret schemes.

 

[sfx2000]

what about these NAS-specific drives where the rotation speed varies (I suppose by track location)? Seems WD red et al are taking this to trade secret schemes.

Just IM'ed a friend of mine over at WD - "not true, the drives spin at a specific RPM, however, from capacity to capacity, they do spin at different speeds, but within a single drive, the RPM's are constant."

WD Red's have firmware that is optimized for NAS type of applications.. same goes with the other vendors. Which makes sense.

 

[stevech]

Just IM'ed a friend of mine over at WD - "not true, the drives spin at a specific RPM, however, from capacity to capacity, they do spin at different speeds, but within a single drive, the RPM's are constant."

Last WD Red 3TB drives I bought go to great extremes to NOT say the rotational speed. They touch the subject with very fuzzy words. I did not like that.

Going green and all that.

 

[sfx2000]

Last WD Red 3TB drives I bought go to great extremes to NOT say the rotational speed. They touch the subject with very fuzzy words. I did not like that.

Yes, and that's what causes some of the confusion - a 3TB will spin at a different spindle speed than perhaps a 1TB would, or a 6TB.

 

[stevech]

Yes, and that's what causes some of the confusion - a 3TB will spin at a different spindle speed than perhaps a 1TB would, or a 6TB.

but perhaps not one of the traditional 5400/7200 RPM speeds?

 

[RMerlin]

but perhaps not one of the traditional 5400/7200 RPM speeds?

I forgot which drive that was, but I know that one of these "stealth RPM models" was rotating at 5800 rpm.

 

[sfx2000]

I know Seagate consumer NAS drives spin at 5900RPM...

 

[stevech]

Why are manufacturers so secretive about rotational speed now? Trade secrets to be best-green? Assuming they use a constant speed for any given drive model.

Or is it that to print a brief specification is too costly in ink?

 

[sfx2000]

Why are manufacturers so secretive about rotational speed now? Trade secrets to be best-green? Assuming they use a constant speed for any given drive model.

Or is it that to print a brief specification is too costly in ink?

Well - for me it about bit-rates across the bus and seek-times - I really don't about about how fast the spindle spins if the heads are fast and the firmware can carry the bits across the bus...

And yah, probably in spinning disk land, trade secrets are more important than ever, as they're getting hit hard by developments in SSD's - which R&D might cost more, but production costs much less over number of units...

solid state always wins in a cost battle...