Which Enterprise-grade HDDs for a RAID5 NAS ?

British

n00b
Joined
Nov 30, 2010
Messages
14
Hello,

I'm going to buy a NAS quite soon (Synology DS1010+), but I still don't know what to put inside.
It will run 24/7, and I'd like to use 2TB Enterprise HDDs for (hopefully) increased reliability.

By doing some research, I found only WD Caviar RE4 and Seagate Constellation ES that are labeled as such.

My concern is that "regular" WD drives have (at least) that WDIDLE issue, and Seagate... well, I don't really remember what problem they might have (heat/noise, or is it Hitachi ?).

Do the Enterprise ones have the same issues ?
What would you recommend (any experience would be appreciated) ?
 
Hitachi has the UltraStar A7K2000 series.

I've mainly used Hitachi and Seagate. Both are fine (at least speaking in terms of enterprise). The new Constellation ES were designed from scratch where as the old Barracuda / Barracuda ES.2 were based off the 7200.XX line.

I looked at the Synology Hard Drive Compatibility list and the 2TB Seagate is on that list, but I don't see the Hitachi. So perhaps look at the Seagate.
 
The WD RE drives are very nice drives, probably what I would go with if I needed "Enterprise" drives.
 
I would put Samsung F4's inside, provided that NAS does not need TLER-enabled disks. TLER is normally the prime reason for buying 'RAID edition' harddrives, which are physically the same as normal consumer harddrives but alledgedly tested better and have different firmware. I don't know if it's worth paying alot of money for that, especially if you do not need the TLER feature. In that case, just buy cheap F4s. Also saves you with power and heat.

Keep an eye on the LCC (load cycle count) in SMART data.
 
Western Digital RE4s. But once you go beyond 500GB it gets very pricey.
 
Thanks for the input guys.

The reason I'm thinking about Enterprise-grade drives is because they are supposed to be much more reliable than desktop models (even notwithstanding TLER): better MTBF and whatnot.

I'm not 100% sure of what TLER means, english not being my mother tongue (ah !).
This a concept I have somewhat of a hard time wrapping my mind around, but this is how I understand it:
This enables drives to have a shorter response time (to... something ?! regular requests or something else ?), because RAID systems tend to drop out drives that take to long to respond.

I remember reading discussions about TLER being used only on hardware RAID, and thus software RAIDs not making use of it.

Yet another thing I'm not sure about is if software RAIDs (such as Synology's) might drop drives for not responding in a timely fashion as well.
Logic tells me it should not, and so the software should not drop drives for this (or even something else besides "real" problems ?), but I'm still wondering.

Besides enterprise drives, the Hitachi Deskstar seems indeed nice, even if there are people reporting it is noisy and hot.
It's even in the DS-1010+'s Compatibility List (along with the UltraStar).

PS: please feel free to correct anything stupid or totally inaccurate I might have said :]
 
There's a TLER thread on this forum i think, you may want to read it.

You're not far off though. TLER applies to how long harddrives may perform 'recovery' when they encounter a sector they cannot read very well, if at all. Normally, the disk would continue trying to recover it, up until 120 seconds or 2 minutes since it first encountered problems with that sector. It could be that after 2 seconds it can read that sector again or it could take 70 seconds, or it may just be unreadable. When the drive 'gives up', it returns an I/O error, telling the controller that it cannot read that sector.

The problem is that RAID controllers cannot distinguish between a totally failed drive, and a drive performing recovery and just not responding for some time. The usual behavior is to 'fail' that disk entirely after it does not respond for 10 seconds or so.

What TLER (Time-Limited Error Recovery) does is limit the default max 120 seconds recovery time, to a single-digit value; 7 seconds is a common setting. I believe 10 seconds is where the RAID controller considers that drive to be failed. So by limiting to 7 seconds, you make sure the RAID controller never times out the drive, but rather that it receives the I/O error.

All depends on how your hardware RAID implements this, but TLER also applies to software RAID and even single disks.

TLER is also dangerous, because you limit the ability for the harddrive to recover your data. If you lost your redundancy, or using TLER on non-redundant configurations (JBOD, RAID0, bare disk, degraded RAID5, double-degraded RAID6) you effectively restrict/limit/kill your last ability to recover that data, which may not be what you want.

On non-Windows software RAIDs, TLER is not required and best kept off depending on what exact software you use. On Hardware RAIDs, TLER is still 'necessary evil' to prevent something simple as a weak/bad sector from being kicked out of the array.

I'm not sure how those ready-made NAS like Synology work; they should tell you whether you need TLER-capable disks.
 
The reason I'm thinking about Enterprise-grade drives is because they are supposed to be much more reliable than desktop models (even notwithstanding TLER): better MTBF and whatnot.

im not sure if that is generally true. at the least- i have not seen any data supporting this statement. i would say there is a big overlapping grey area between the high end "consumer" models and the "enterprise" models in terms of reliability. i do not think that just because you spend +40% more for an "enterprise" drive, that it will automatically outlast a "consumer" drive in most situations.

some specific models are worse than others though...
 
I have seen studies that say there is no difference in reliability between the "enterprise" grade drives and consumer grade drives. Both will still have the expected > 1% annual failure rate.
 
Real enterprise drives (SAS 10k rpm+) are built differently, with more sturdy components that are tested with 100% seek rate (512B random reads). Those justifiably have higher MTBF values. But you also have the 'RAID edition' consumer drives, which come from the same manufacturing line and are just equiped with different firmware and supposedly tested better, though i think the difference is very minor. Seems more like an easy way to get a bit extra margin selling firmware (which doesnt cost the company any extra) for people who need the TLER feature.

Most RAID edition drives and enterprise drives have extended warranty, though. This would be the most useful and perhaps make it very worthwhile to pay for it, but only if the difference is small enough to justify.
 
Thanks for the explanation on TLER, sub.mesa, that was quite interesting.

BTW, I found the article about TLER that I've had read a while back: http://www.smallnetbuilder.com/nas/nas-features/31202-should-you-use-tler-drives-in-your-raid-nas

Here the bottom-line:
So is there any benefit to using TLER / CCTL / ERC drives? Maybe. These features usually come on "Enterprise" grade drives (WD Caviar RE series, Seagate Barracuda ES, ES.2, Samsung Spinpoint F1), which are built to take the constant, hard use of business environments. So investing in these more expensive drives is probably a smart move if your NAS is under constant heavy use. But it will be the more robust drive construction and not TLER / CCTL / ERC that will make your RAID NAS more reliable.

A little anecdote, for the record: I have a Synology CS-406e that I bought 3.5 years ago.
I filled it with Seagate ES drives at the time.
Now a few weeks ago, one of the drives died on me, but thanks to buying enterprise drives it's still under warranty :D
It was anecdotic, so don't draw any conclusion out of this ;)

I'm probably leaning towards the Hitachi DeskStar, BTW, the UltraStar costing a kidney or two.
Of course if one dies on me after "regular" warranty is off, I'll be a sad panda, but it'd probably be most cost-effective in the short run (maybe even in the long one).

I'm steering away WD and Seagate, for different reasons that tingles my paranoia (and which can be found on about any hardware forum around the internet).

Nothing is set in stone yet, though.
 
Back
Top