DIY server: Good SATA HDDs

Yup I have used lots (probably 16 or so) of the WD "Red" drives in NAS applications and not had any failures as yet after several years

So they are certainly a "safe" choice. Performance is ok - you need gig ethernet to keep up with them.

I have used the 1TB version (Seagate 3.5, 1TB, SATA3 SSHD Hybrid Desktop Hard Drive, 7200RPM, 64MB Cache, model ST1000DX001).

Again it seems like a nice drive - I have not used enough of them for long enough to comment on reliability.

(note the one I have used is probably not certified for NAS/Server style

24/7 operation in a box with other drives)

Performance is impressive in a desktop environment - no far off the speed of pure SSD but without the cost of the hassle of needing to deal with different partitions for different purposes (not a problem for me, but for some users, coping with more than one partition is a non starter!)

Reply to
John Rumm
Loading thread data ...

I find its unwise to use a blanket brand vs brand comparison... its the detail of the individual device that makes the difference.

For example, I used a pair f 2TB Hitachi "Green" drives in a NAS application once. They seemed reliable enough, until you realised that due to a feature of their firmware, they decided that after 8 seconds of no access they would unload the drive heads. In an application like this it basically meant they spent all day loading and unloading the heads. So in the space of 16 months they romped through their maximum load/unload count of some 340,000 cycles, and started failing SMART tests. (those particular drives are still running in a gash machine - but I would not trust them for anything that matters now)

Reply to
John Rumm

How do you check if these are authorized resellers?

formatting link

Reply to
dennis

Worth noting that the figures are from Backblaze[1] - and they have a rather different setup to many outfits in they way they use (and in some cases source (its worth reading their blog posts on the big drive shortage of a couple of years back) their drives).

[1] Online backup and archiving operation using self built servers optimised for low cost per GB over all other factors)

I have used some of the cheapish Kingston ones, and they seem ok so far.

Reply to
John Rumm

In the interests of mixing drives, are there any other makes that fall into the same space? Plenty of enterprise drives, but WD Reds seem unique.

Yep - been there for years now :) The internal NIC in the HP Gen 7 Microserver apparantly drives at 900ishMbit/sec so that's fine too.

Interesting. I could be tempted to take a punt...

Cheers!

Tim

Reply to
Tim Watts

Sounds like a true "green" product. Nice idea but completely useless!

Reply to
Tim Watts

How fast are they? Fast enough to saturate a SATA-II interface? That's my laptop's benchmark because the Sandisk does indeed become constrained by the Lenovo's SATA-II port. Still bloody impressive speed though - knocks spinning metal into the long grass both on raw throughput and of course random IO.

Reply to
Tim Watts

I don't have benchmark figures, but can give you some real world examples...

Its a Kingston SVP200S360G on a ASUS P8Z68-V LX mobo with 6Gb/s SATA

Load any component of MS Office 2010 - approx 1.5 sec

Photoshop PS CS5.1 about 3 - 4 sec

Copy 1.2GB file over network from the NAS to the desktop folder - approx

32MB/sec for 40 secs (limited by NAS throughput)

Duplicate file locally - interesting as you can see the cache on the drive come into play - it starts off at over 115MB/sec, but falls to about 42MB/sec toward the end of the copy. Total time about 15 sec

Repeat that with a 150MB file however and the duplication time is near enough a second or so.

Without a benchmark its hard to assess read speeds, but I would guess they are up in the 400 MB/sec range - so probably enough to saturate a

3Gb/s SATA.
Reply to
John Rumm

+1
Reply to
Johny B Good

Before you decide on Seagates, take a look at this:

It's a bit of an eye opener (which, incidently, confirmed my prejudice against Seagate).

Reply to
Johny B Good

Yes - I did read that - it is interesting.

I did mention that my work's SAN is full of ES grade Seagate SATA disks and that's not done too bad - 3 disks in 4 years out of 48 total.

Reply to
Tim Watts

Yep, crucial SSDs are brilliant, apart from the the V4. Big mistake with that, we sent loads back. Wierd random hangs. In the end they sent us M4s as replacements for no extra cost. Solid drives, as are the M500s

V4 is almost completely irradicated from their website now, hardly any mention of it. I suspect they seriousely regret it - really dented their reputation!

M550 I've no experience of yet, but I'm sure we will.

Darren

Reply to
D.M.Chapman

That's pretty good. I guess these are 7k2 drives? Our primary array is currently 345 disks, and we lose 2-3 a month at almost 5 years old. Around 10 to 12 thousand IOPs on average so worked fairly hard.

That's 15k SAS drives though, they run hot. HDS ship a combination of drives to try to avoid a duff model causing problems. Can't say I've noticed a pattern. All manufacturers have good and bad spells it seems.

Best to just assume they are going to fail and plan for it, then you might be pleasantly surprised. WD green drives can be a real pain as they are rather too keen to go to sleep...

ZFS is cool, but avoid filling up the FS.

formatting link

formatting link

Under 80% you'll be fine. Above that, oddness can happen :-)

Darren

Reply to
D.M.Chapman

Indeed, it's the random IO that makes the difference. Most users will never notice the difference in transfer rate between different SSDs. The seek time (or rather lack of!) is a big win over spinning rust

Darren

Reply to
D.M.Chapman

Yes. It's gets the IOPS through brute force spindle count - and not buggering it up with crap controllers (saw that once - not EqualLogic).

Hmm - might give that a miss then. I can cope with the old adage that a near full FS suffers slowness and bad fragmentation, but "oddness" at

80%??? I'll play with it during setup for the knowledge, but I'll think I'll go back to good old md-raid5 which is a nice well behaved beast :)
Reply to
Tim Watts

My laptop Sandisk boots *very* fast - according to bonnie++, it can sequential-block-read at 271MB/s and sequential-block-write at 254MB/s

That's pretty fast!

Reply to
Tim Watts

Full results for anyone who's interested:

(fs was XFS on top of LVM)

squidward # bonnie++ -u root -f -d /test/ Using uid:0, gid:0. Writing intelligently...done Rewriting...done Reading intelligently...done start 'em...done...done...done...done...done... Create files in sequential order...done. Stat files in sequential order...done. Delete files in sequential order...done. Create files in random order...done. Stat files in random order...done. Delete files in random order...done. Version 1.97 ------Sequential Output------ --Sequential Input-

--Random- Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block--

--Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP squidward 11528M 254115 21 125937 27 271298 32 6254 233 Latency 18280us 224ms 4459us

7133us Version 1.97 ------Sequential Create------ --------Random Create-------- squidward -Create-- --Read--- -Delete-- -Create-- --Read---

-Delete-- files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP 16 28205 68 +++++ +++ +++++ +++ 29605 73 +++++ +++

+++++ +++ Latency 4637us 271us 155us 287us 77us 185us 1.97,1.97,squidward,1,1395353201,11528M,,,,254115,21,125937,27,,,271298,32,6254,233,16,,,,,28205,68,+++++,+++,+++++,+++,29605,73,+++++,+++,+++++,+++,,18280us,224ms,,4459us,7133us,4637us,271us,155us,287us,77us,185us
Reply to
Tim Watts

Something else worth noting is that enterprise drives are in one sense designed to fail more often i.e. they keep track of additional performance heuristics that can cause them to enter a soft error state when potential faults are detected. The "soft" failures usually leave the operator aware there is a brewing problem, but at a point where the data are recoverable. With the consumer drives the problem may only be detected with a hard failure.

Reply to
John Rumm

I've used ServerPlus for many purchases, including two microservers with cashback, never any problem and they are helpful if you need to phone them etc.

Reply to
djc

indeed - we have a few a month fail in our main disk array. I'd say around

2 thirds of them are preemtive failures. We had a spell of over temperature bearing faults throwing disks, but a controller firmware fixed that (1).

The drives spare out by copying data off the failing disk onto a spare and then the controller takes them offline.

Less frequently one just dies and vanishes - more expensive to rebuild then.

Darren

(1) the new firmware just increased the temp that flags a fault :-)

Reply to
D.M.Chapman

HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.