Which Hard Drive is Best for Your Home Server or NAS?

Earlier this summer, I published We Got Served’s review of Western Digital’s brand-new WD Red 6TB NAS hard drive. As I mentioned in the feature, both WD and Seagate have been quick to niche their product line-ups, tailoring their core storage technologies with tweaks and value-added features to serve a wide variety of consumers.

In the course of writing the review, I was struck by the fact that for the average home server/NAS owner, selecting the right storage for your rig is more complicated that it ever has been. In the past, it was easy – you had consumer desktop drives and enterprise drives. Given the scarily-high prices of the latter, it was a simple choice.

But then came the so-called “green” drives – desktop drives with a low-power twist. Then the aforementioned NAS drives, “pro” drives for small business, drives for enterprises, drives for data centers, drives for surveillance equipment. Each with SATA 3 Gb/s and 6 Gb/s options. Sure, you could just go for the ones that say “NAS” right? That would save you a lot of time worrying about selecting the right drive, wouldn’t it? As the manufacturers have done the worrying for you!

But… really? Should you be paying a price premium for NAS storage, compared to standard desktop drives? Do those value-added features truly add value? And what are all those features anyway? Perhaps you shouldn’t be trusting your precious data to a consumer drive at all and invest in expensive small biz or enterprise drives instead? Surely the more you pay, the better the drive?

In this (reasonable exhaustive) feature, we’ll try out a swathe of hard drives on a number of test devices to compare how they perform. We’ll look at some new drives and as we know that many of our readers like to re-purpose old machines and older drives for their home servers, we’ve dug out a couple of old drives we had knocking around to add them to the mix too.

We’ll explore some of the features that the manufacturers market to convince you that their drives really are the right ones for you. Whilst I doubt very much that we’ll settle on a single model that we believe is right for every user and every server, in every situation, hopefully our exploration will help you on your own journey to find your chosen model.

Having spent the last seven years reading and writing about network attached storage, my promise to to make this comparison as simple and clear as possible. There will be benchmarks, but I’m not going to dive into every comparable stat imaginable – I’ll leave that to the more technical sites out there that really excel at the hardcore detail. Rather, I’ll look at the indicators that I think are most important in determining NAS performance – there are only a few, so no-one should be bamboozled by stats.

With that all said, let’s introduce our test subjects.

Our Test Hosts

I thought it would be useful to compare performance on a range of consumer and prosumer/small business devices – brand selection was based more on what devices I had available here at WGS rather than any specific intent, but I wanted to ensure we had a representative selection of devices released over the last couple of years.

So, you’ll find our test hosts include a basic, twin-bay device with a budget processor, a four-bay Intel Atom-based NAS, an Intel Core i5 Windows 8.1 PC and a rackmount ASUSTOR AS-609RS.

One important consideration for you when selecting your NAS or home server is the capacity of drives that a particular device supports. As we move into an era of huge hard drive capacities – 5 TB, 6 TB and beyond – you may find there’s a limitation on the maximum size of RAID array that a device supports, particularly at the lower end of the market. For example, you may assume that all modern four-bay NAS devices would support arrays of 24 TB (four x 6 TB drives), right? Wrong. Let’s pick a device such as ASUSTOR’s AS-304T – a reasonably new NAS device. Dig into the device’s specifications on the web, and you’ll see “due to system kernel limitations, the maximum capacity of a single RAID volume for this model is 16TB.” Obviously, other devices from ASUSTOR and others support arrays well in excess of 16 TB, but not all do.

So, while we’ll test drive sizes up to 6 TB on our test devices, do make sure to check any capacity limitations on your devices before deciding on your own choice of drives.

The Test Setup

To test performance, two of each hard drive model were installed in the NAS device/server and configured in a RAID 1 (mirrored) array. The server was connected to a production, real-world Gigabit network (no polished lab conditions here!) with an Apple Time Capsule acting as the router. We then measured network file transfers using an Intel NUC as the client PC. This device is powered by an Intel Core i5 processor and was running Windows 8.1 as the operating system.

The Hard Drives

We’ve selected nine drives of varying specification, size and age to benchmark on our test servers. They comprise standard desktop drives you’ll find in your local big-box electrical retailer, specialist NAS drives and drives for small business as well as drives intended for enterprise and datacenter use. Obviously it would be difficult (and dull) to test every drive from every manufacturer in the market, so take this as a representative sample of what’s available in stores (and stored away in cupboards at home).

Here’s the line-up:

For the uninitiated, the table above lists the manufacturer’s brand, model name and SKU (stock keeping unit – think of it as a product code that retailers use). I’ve also included the drive capacity (in terabytes), SATA connection speed (SATA stands for Serial ATA and is rated in gigabits per second – the maximum speed at which data is transferred through the hard drive’s physical connection to the motherboard), Cache size (embedded memory that’s used as a data buffer between the physical disk platters in your hard drive and the home server/NAS – rated in megabytes) and Spin speed (the speed of the drive’s disk rotation – measured in revolutions per minute). Faster rotation doesn’t necessarily mean faster transfer rates though and as you can imagine, the faster you spin a disk, the hotter it can get and the more power it may use.

SATA connection speed, Cache size and Spin speed can all have an influence on data transfer rates, so are generally used by manufacturers to classify their products at a high level. As we’ll see, there are many more features built into these drives that may also influence performance and reliability.

Let’s talk through the list by category.

Desktop Drives

We’ve picked out two desktop drives for our round-up. The  has been around for some time now (in fact, the company no longer use the well-known Barracuda product brand – it’s just Seagate Desktop Hard Drive) and I’m sure that most of you will have bumped into one of these drives at some point. It’s a 3 TB drive that connects to the NAS via a 6 Gb/s SATA connection. Like all of the drives on test, it includes a 64 MB cache and rotates at 7200 rpm – that’s generally the maximum speed you’ll find on common desktop hard drives.

seagate-desktop

At the time of writing, the drive is available in capacities up to 5 TB with the largest drive benefitting from a class-leading 128 MB cache. I’d categorise our selection as a “mid-range” drive.

In the Desktop category, the Seagate Hard Drive competes with Western Digital’s WD Blue product brand, and we’ve included the 1 TB WD Blue WD10EZEX1 TB WD Blue WD10EZEX in our test. Like the Seagate drive, it benefits from the same 6 Gb/s SATA connection, 64 MB cache and 7200 rpm spin speed. However, while Seagate’s standard desktop category extends all the way to 5 TB capacity drives, Western Digital caps their Blue range at 1TB. For higher capacities, you’ll need to look at WD Green drives. So let’s do that.

Low Power Desktop Drives

Western Digital’s WD Green hard drives range from 500 GB through to a massive 6 TB. They’ve historically been positioned as “cool and quiet” drives, with lower thermal output than other hard drives – the thinking goes that, over time, additional heat can translate into reduced drive reliability, increases in power consumption and additional noise (your server’s fans have to work harder to extract all of that heat, right?) It’s here that we start to see Western Digital add features into their drives to woo perspective buyers.

Total
15
Shares

25 comments

    1. That’s a great link! Just remember that those guys are in a datacenter and you may not be hammering your drives as hard at home or in your small business! But all things being equal, the reliability comparison between the drives they’re using is interesting.

  1. Hi,

    WD Re drives spin at 7200rpm, not 5,400

    if you actually click the link in your own paragraph you will be taken to an amazon page where the title proclaims the Re to be 7,200rpm

    “Well, if you check back in our specification table, you’ll see that the WD Re drives spin mire slowly than the Se drives – 5400 RPM vs 7200 RPM and that translates into lower power consumption. ”

    I suspect the confusion is because you have looked up the specs sheet for the old RE4-GP drives, which are NOT part of the same series as their current model Re/Se/Xe enterprise drives (one is RE , one is Re .. isnt that clear), but were in fact raid-optimised GP (green power) drives

    here’s the Re specs page on the WD site, clearly showing every single Re drive spins at 7,200 rpm

    http://www.wdc.com/en/products/products.aspx?id=580#Tab3

  2. Those Seagate Barracuda Desktop drives are no longer on the QNAP compatibility lists, as far as I can tell.

  3. Thanks Mr. Walsh for a great article.
    In my opinion the more expensive drives would be worth it if their guarantee was longer.
    Say 3 years for a desktop drive, 6 years for a NAS-drive and 10 years for a Enterprise drive.

  4. Hi Terry: I did my own review of drive reviews and wound up picking four Western Digital Red 4 TB drives for my home server. Unfortunately I intended to use WHS 2011 and 4TB drives are not supported for server backup so instead I’ve installed Windows Server 2012 r2. I was able to use the RAID5 array built with WHS 2011, interestingly. There are drive reliability results from an increasing list of other writers/publishers that agree on the following: Western Digital is making the most reliable drives now. WD also now owns HGST which was formerly IBM and those drives, according to reviewers, appear to be enterprise-class drives packaged and sold as consumer drives. They are the most reliable but also count now as WD drives. Eric

  5. “equating to a 35% MTBF improvement. MBTF? Mean Time Before Failure”

    Direct quote from your review.

    Dude. Once, even twice I can ignore it. SEVEN TIMES. Seven times you made the typo, even in the expansion of the actual acronym. Every time I stumbled across it, I kept thinking, “WTF is MBTF? mean between time failures? nawww that doesn’t make sense. Maybe the manufacturers came up with a new buzz word…nnnnaaawwww….”

    Sorry, for this, but it’s so I feel better. MTBF MTBF MTBF MTBF MTBF MTBF MTBF.

  6. I’ll be buying a QNAP TVS-871 when they come out next month. This article was well written and confirmed what I was already leaning towards. I’ll be replacing a 10 year old HAMMER drive (with two 1 TB drives mirrored). The funny thing is the HAMMER drive is still running like a champ but with videos and family photos growing I’d rather setup RAID 6 to protect these things.

  7. Very interesting options… I am looking into this just stay have independence with all of my own data. I wonder if this might become the next major wave for home computing? Everybody keeping their own data on their own servers. Sharing it with “Torrent style” software so as to have more control, instead of being at the mercy of social media and other aggregators. Heck of a market growth potential for the server hardware guys if the software is there.

  8. thinking about the pratically identical speed of all the NAS and drives you tested…
    didn’t read all the article, but I think you didn’t mentioned the speed of the net where you tested the devices…
    Gigabit ethernet “should” provide a maximum theoretical speed of 125MB/s…real speed is less than that…
    so, what I think is that all the devices combination you tested, probably had different results…for example, some of the NAS used have double Gbit port, maybe trying to do speed test with different PC over the same network with 2Gbit port connected the results are different and permit to understand which couple (drive/nas) is faster than another. The only useful information I can see is that speed slower than 90MB/s determine which drive or nas are worst than the others…

      1. I wasn’t obviously talking about internet…local network has its maximum speed too! (or you think wires at home are faster than the star trek ship?) and, it’s funny, it’s almost the same maximum speed you measured on all the devices tested!!! I want to do an example like yours of the focus vs. the ferrari…saying that all the devices go to 100MB/s is like saying that 130km/h is the maximum speed of both a focus and a ferrari just because they’re respecting the speed limit on the highway.
        A confirm of this is the little note that QNAP add to some of their speed test, that says that they tested a NAS using 2PC simultaneously to reach speed of about 200/250 MB/s

        1. The only way to get an absolute speed test out of a set of drives is to
          pull the data on a 10GB network so that the network is not s speed
          limiting factor and to ensure that the system transferring data for the
          test is faster than the spinning disks being tested by using an array of
          SSDs. The trouble is, at some point, you will probably still be
          limited by a hardware speed limit.
          I have a set of 10 Western Digital Re drives that will push enough data to flood even a 10GB card with all it can take.

  9. Outstanding article. I had gotten lucky and selected a lot of 4Tb WD Red NAS HDD units for 3 Synology NAS units, and am very pleased with the performance and reliability. On a 4th Synology unit I ended up with some Seagate NAS class HDD units, and in fairness I have had no problems with them. The WD Red HDDs do run a little cooler than the Seagate HDDs, per the Synology reporting interface.

  10. I am building a home NAS system. I will run Unraid as my OS. I would like to use a WD Black desktop 1tb drive as my cache drive. Would that be a good choice or should I use an actual WD NAS RED instead? I plan to use 4 1tb WD Red drives for the actual storage, 1 2tb WD RED for Parity and as I said 1 WD Black desktop drive as my cache. What are you thoughts on this idea?

    1. Without knowing why you need cache, my inclination would be to use more real memory or SSD for cache and save the 3 1/2″ bay for additional storage or later expansion space, if you can find a place for 2 1/2″ SSD that doesn’t fill a disk bay.

      1. Thank you. I will be using an SSD as a cache drive within the NAS system. That leads to an additional question: Since I am using 4ea 1TB drives for storage, 1ea 2TB for parity is there a size limit to using an SSD as a cache drive? I was thinking of using a 240gb SSD but if a larger or an additional 240 drive would be better than that is what I will do.

  11. Hi,

    Interesting read. It would be great if future update of this article also include noise measurements and more recent drives.
    In my case, I noticed that some older samsung HD103UJ are much noiser than my WD 30EZRX. I wish I had done better research back then.

Leave a Reply