25

I've got an application which requires data recording in a outdoor environment, and I am interested in the reliability of SSDs vs HDD when placed in a cold (down to -20) and hot (+50) ambient environments. Intuition leads me to believe SSDs will be more reliable, with the possible exception of high temperatures. Air conditioning enclosures is not an option.

Does anyone have any information on disk reliability in these situations?

John
  • 353
  • 1
  • 3
  • 8
  • 7
    Have you checked the datasheets of any specific drives? There are industrial HDDs made for wider temperature ranges. – mulaz Feb 18 '14 at 14:52
  • 1
    possible duplicate of [SSD head / cylinder / cluster details](http://serverfault.com/questions/505929/ssd-head-cylinder-cluster-details) – ewwhite Feb 18 '14 at 15:14
  • 1
    How much data are you logging/how heavy is the OS? If performance isn't really an issue, I would look into using an SD/CF card or maybe even a USB flash drive. – Nick T Feb 18 '14 at 18:16

3 Answers3

26

Look for an industrial or ruggedized SSD for this application.

A good example of a proper product spec.
http://www.pretec.com/products/ssd-series/item/sata-ssd-series/a5000-industrial-grade

．Standard 2.5" SATA III SSD, compatible with SATA III/II/I interface
．Capacity: 32GB ~ 256GB
．Data transfer rate: Up to 490 MB/s
．Built-in ECC (Error Correction Code) function
．Support ATA-8 command and SMART function
．Temperature
I.  Operating Temperature: 0℃ ~ +70℃
II. Extended Temperature: -40℃ ~ +85℃
III. Storage Temperature: -55℃ ~ +95℃
Community
  • 1
ewwhite
  • 197,159
  • 92
  • 443
  • 809
  • Hey thanks for the links- Do you know what is the difference between operating temperature and extended temperature? – John Feb 18 '14 at 15:26
  • 6
    absolutely. just like any "spike" in a metric, systems can usually handle that spike up to some extended period of time before accelerated breakdown. bottom line is you do not want your extended temp to be your normal operating conditions. and don't be fooled by people telling you they have some device that functions normally in extended conditions. there will always be outliers. also, you run the risk of voiding mfg warranty (but not something i would get worked up over) – au_stan Feb 18 '14 at 15:43
  • Do you know if these temperatures are for ambient temps, or working temps? Living in a hot country, I've had individual electronic components (mainly electrolytic caps, of which there wouldn't be any in an SSD) that shit themselves with extended 40℃ ambient temps, even though they were rated to 115℃. – Mark Henderson Feb 18 '14 at 23:36
  • @MarkHenderson Working temperatures... I have industrial SSDs in some of the packing and sorting machinery located in the coolers/freezers of warehousing environments. – ewwhite Feb 18 '14 at 23:51
  • A typical non-industrial SSD ([Intel 530](http://www.intel.com/content/dam/www/public/us/en/documents/product-specifications/ssd-530-sata-specification.pdf)) is rated for the same temperature ranges. – Martin Schröder Feb 19 '14 at 19:17
  • 1
    It's not just about temperature... – ewwhite Feb 19 '14 at 19:18
-3

The main difference of HDD and SSD is the Storage Capacity and its Related Costs. You can buy SSD supported devices with storage of up to 4 terrabytes, but they come at a cost.An HDD, meanwhile, will give you storage capacity of up to 2 terrabytes on a laptop and 10 terabytes on a desktop.

Dustin DeTorres
  • 1
  • 2
  • 1
    This is a very poor 'answer', it doesn't address the question in any way and in fact simply ensures that all of the sysadmins here - your future customers don't forget - see how badly you understand simple things and thus will never do business with you - bravo. – Chopper3 Apr 12 '17 at 16:34
  • 1
    If you link to your website, you need to state that explicitly in your post. Please read [How not to be a spammer](/help/promotion). – Glorfindel Apr 12 '17 at 16:35
  • I don't want to spam, I deleted that post. Thanks for the information though. – Dustin DeTorres Apr 12 '17 at 16:36
  • 3
    @Chopper3 please respect the be nice policy of SE (and I saw that comment earlier) – CalvT Apr 12 '17 at 16:39
  • And his account on [SU] was spamming his blog for a different Brisbane based data recovery company. – DavidPostill Apr 12 '17 at 16:46
-4

Yes, there are chips which can work on extreme temperatures (both in cold or warm), but they are much costly. In practice, in most situations it is better (a LOT cheaper) to solve the problem of the temperature stabilisation as to use such electronics.

The temperature between -20 - +50 C isn't really extreme, especially in case of always working hardware, because its normal ohmical heat will him enough for the normal work. For the higher temp I suggest to use simply air conditioning.

And yes, SSDs are more reliable because they don't contain moving parts.

Another solution (I don't know it were applicable in your situation): you could use a simple, minimal embedded hardware which stores its data in a better, protected area, and reaches it with cable network or with wifi.

peterh
  • 4,953
  • 13
  • 30
  • 44
  • 2
    this answer offers very little. examples? useful products? experiences and proof to back up your claim that SSDs are more reliable? – au_stan Feb 18 '14 at 15:46
  • 1
    @PeterHorvath it is *generally expected* that if you make an assertion (as you are doing in your answer: that ruggedized chips/devices are "much more costly" than standard ones) you will back it up with a reference. There's no "directive" on this as it's not a stack exchange policy - it's just *common sense*. You are of course free to disregard common sense, but answers with appropriate references tend to get more upvotes than ones that just make an assertion. – voretaq7 Feb 18 '14 at 16:46
  • @NathanC If somebody thinks different, for example thinks that "yes, coldresistant chips are very cheap", I could come out a lot more and weigher proof behind my opinion as he. But I didn't happened, maybe nobody thinks it so. – peterh Feb 18 '14 at 16:46
  • 1
    @PeterHorvath most ICs are fine in the cold, see [here](http://electronics.stackexchange.com/a/6980/1407). I find it astonishing that you think purchasing, installing, and powering an AC unit is cheaper than buying perhaps a few ruggedized disk drives. – Nick T Feb 18 '14 at 18:11
  • 1
    Awful answer. Adds no value, misinforms. -1. If it truly was easier to solve the temperature problem, people would. But clearly, that's not the case because a market exists for rugged hardware. You really think it's easier to cool down an assembly line than install rugged hardware? What about computer hardware that, say, lives its life near an internal combustion engine? (You know, like the ones in your car.) How about computers that need to go outside? It might not get cold where you live, but a lot of the world does, and don't want computers that they can't use during the winter months. – HopelessN00b Feb 18 '14 at 19:32
  • @HopelessN00b What I said is that the easiest way to make temperature-resistant system to use heating and temperature conditioning. It is the hard reality - if you have a better idea, you should write this as an alternative answer. And I can't interpret your "questions", and also not your real problem. Maybe a little bit of successful hardware-related projects could help you a lot. – peterh Feb 19 '14 at 12:21
  • @PeterHorvath The better answer is ewwhite's - use a rugged/industrial grade system, rather than trying to climate control everywhere that might need electronics installed. Such as, for example, under the hood of a car (near that hot internal- combustion engine), on an assembly line, or outdoors. There are plenty of places that it's simply not practical to climate control, and someone wants to put a computing device. Rugged or industrial-grade computer systems are the solution that's designed for exactly that use-case. – HopelessN00b Feb 19 '14 at 12:44
  • @HopelessN00b No, but heat-resistant chips are coming in small quantity and thus they have a lot bigger price. Complex hardware practically can't be made from them. Even the billion-dollar satellites are working with thermal stabilisation, even the radioactivity sensors of the nuclear plants are working with thermal stabilisation and not with heat-resistant chips. Yes, enwwhite's answer is better (also I upvoted it), but it is irrelevant in the context of _our_ arguments. Temperature control doesn't mean always moving parts (see peltier elements), although in practice it is mostly so. – peterh Feb 19 '14 at 12:51
  • @PeterHorvath ...all of which seems pretty irrelevant in the context of professional systems administration. We don't have access to semi-conductor fabrication facilities, and can't manufacture thermal stabilization technologies into the systems we deploy, so our only real recourse is to buy industrial-grade systems or components. And, just to respond to the academic argument piece, the problem with low temperature and electronics is condensation. So satellites don't need thermal stabilization for their chips, because there's no water vapor to condense in the vacuum of space. – HopelessN00b Feb 19 '14 at 14:02
  • @HopelessN00b Professional system administration doesn't mean always remote management of servers in stabilized serverfarm environments. IMHO you had to pleased to outer, hardware-related knowledge. This unjust mobbing made me really worry. – peterh Feb 19 '14 at 14:25
  • @PeterHorvath The point isn't that we're against hardware engineering information, the point is that it's not helpful/useful in the context of this question or problem. The guy's asking how to do data recording in an outdoor environment. Telling him it's possible fabricate/engineer integrated chips to withstand extreme environments is **not** a useful answer for a sysadmin. (Which is the tooltip beside the down arrow). Telling him he can purchase industrial-grade components **is** a useful answer. If this was hardware-engineering.SE, your answer would be useful, but it isn't, so it's not. – HopelessN00b Feb 19 '14 at 14:40
  • @HopelessN00b No, it is what you like better, because it is not about doing something, but to buy something. I don't think a sysadm should buy (or let buy) things and don't have to do anything with the hardware. – peterh Feb 19 '14 at 14:54