Hard Drive Temperature – Does It Matter?

May 12th, 2014

blog-drive-temperature

How much does operating temperature affect the failure rates of disk drives? Not much.

The unlimited online backup service provided by Backblaze requires a lot of storage. In fact, we recently passed the 100-petabyte mark in our data center. This means we use disk drives. A lot of disk drives.

The Backblaze Storage Pod is designed to provide good airflow over the disk drives, so they don’t get too hot. Still, different locations inside a pod, and different locations within a data center will have different temperatures, and we wondered whether that was a problem for the drives.

What Other People Say

Google and Microsoft have both done studies on disk drive temperature in their data centers. Google found that temperature was not a good predictor of failure, while Microsoft and the University of Virginia found that there was a significant correlation.

Disk drive manufacturers tell Backblaze that in general, it’s a good idea to keep disks cooler so they will last longer.

All Drives: No Correlation

After looking at data on over 34,000 drives, I found that overall there is no correlation between temperature and failure rate.

To check correlations, I used the point-biserial correlation coefficient on drive average temperatures and whether drives failed or not. The result ranges from -1 to 1, with 0 being no correlation, and 1 meaning hot drives always fail.

Correlation of Temperature and Failure: 0.0

Disk Drive Temperature Range

It turns out that different drive models run at different temperatures, and this can throw off the stats when looking at the entire population. If in a given ambient air temperature, drive model A runs warmer than drive B, and drive A fails more, that will make it look like there is a correlation when there isn’t.

This table shows the average temperature, in degrees Celsius, of different drive models:

Model Avg. Temp (C)
Seagate Barracuda LP (ST31500541AS) 21.92
Seagate Desktop HDD.15 (ST4000DM000) 22.10
Seagate Barracuda Green (ST1500DL003) 22.86
Western Digital Red (WDC WD30EFRX) 23.05
Seagate Barracuda LP (ST32000542AS) 23.27
Western Digital Caviar Green (WDC WD30EZRX) 23.46
Seagate Barracuda 7200.14 (ST3000DM001) 24.71
Western Digital Caviar Green (WDC WD10EACS) 25.23
Seagate Barracuda XT (ST33000651AS) 25.40
Hitachi Deskstar 5K4000 (Hitachi HDS5C4040ALE630) 25.42
Seagate Barracuda 7200.11 (ST31500341AS) 25.73
Toshiba DT01ACA Series (TOSHIBA DT01ACA300) 25.82
Hitachi Deskstar 5K3000 (Hitachi HDS5C3030ALA630) 26.46
Hitachi Deskstar 7K3000 (Hitachi HDS723030ALA640) 26.75
HGST Deskstar 7K4000 (HGST HDS724040ALE640) 27.22
Hitachi Deskstar 7K2000 (Hitachi HDS722020ALA330) 27.39
HGST Megascale 4000 (HGST HMS5C4040ALE640) 27.84
Western Digital Caviar Green (WDC WD10EADS) 27.93
Seagate Barracuda XT (ST4000DX000) 30.54

Each Storage Pod in our data center is initially deployed with one model of drive in all 45 slots. It tends to stay that way over time, too, as drives are replaced. Pods with different models of drives are distributed somewhat randomly around the data center, so on the average, each model runs in an environment that is about the same. The temperatures in the table above are due to differences in the disk drives more than differences in their environment.

The first five drives in the above list are all advertised as “green”, low-power drives. It makes sense that they run cooler because they generate less heat.

The chart below shows the distribution of drive temperatures for our four most popular drives. As you can see, all of the drives are well within the 0° (or 5°) to 60° that the manufacturers specify for the drives. And almost all of the drives are in the nice comfortable range from 15° to 30°.

blog-temp-totals

Correlations Between Temperature and Failure for Different Drives

Now, let’s look at the correlation between temperatures and failures for each drive model. Here’s the same set of models, this time sorted by correlation. The correlations that are statistically significant are in bold:

Model Correlation Significant? p-value # dead # alive Avg. Age
(years)
Western Digital Caviar Green
(WDC WD10EACS)
0.18 no 0.07 2 107 4.9
Seagate Barracuda 7200.11
(ST31500341AS)
0.17 YES 0.00 157 628 3.8
Seagate Barracuda LP
(ST31500541AS)
0.12 YES 0.00 195 1992 3.8
Seagate Barracuda Green
(ST1500DL003)
0.05 no 0.61 66 50 0.8
Seagate Barracuda 7200.14
(ST3000DM001)
0.03 YES 0.02 638 4031 1.4
Western Digital Red
(WDC WD30EFRX)
0.02 no 0.67 21 661 0.5
Western Digital Caviar Green
(WDC WD30EZRX)
0.01 no 0.88 22 477 1.7
Hitachi Deskstar 5K4000
(Hitachi HDS5C4040ALE630)
0.00 no 0.82 32 2671 0.8
Seagate Desktop HDD.15
(ST4000DM000)
-0.01 no 0.25 133 9350 0.3
Seagate Barracuda LP
(ST32000542AS)
-0.02 no 0.71 22 363 2.0
Hitachi Deskstar 5K3000
(Hitachi HDS5C3030ALA630)
-0.02 no 0.13 36 4591 1.7
Western Digital Caviar Green
(WDC WD10EADS)
-0.04 no 0.39 21 529 4.4
Hitachi Deskstar 7K2000
(Hitachi HDS722020ALA330)
-0.04 YES 0.01 57 4708 2.9
Seagate Barracuda XT
(ST4000DX000)
-0.04 no 0.56 1 179 0.7
Hitachi Deskstar 7K3000
(Hitachi HDS723030ALA640)
-0.04 no 0.15 14 1022 2.1
Toshiba DT01ACA Series
(TOSHIBA DT01ACA300)
-0.05 no 0.73 2 58 0.7
Seagate Barracuda XT
(ST33000651AS)
-0.05 no 0.35 23 286 2.0

Seagate Barracuda & Barracuda LP 1.5TB Heat Failure

This is the one drive that does show some correlation between temperature and failure rates. The correlations of 0.17 and 0.11 are weak, but they are statistically significant.

It’s interesting that the correlations are similar for the regular 7200 RPM drive and the low-power 5900 RPM drive. The average temperature of the low-power drives is 21.9, while the average for the regular drives is 25.7.

Comparing the failure rates of drives that are below the average temperature for the model, and those above the average temperature, there is a clear difference for these drives:

Annual Failure Rate
Cool (below avg. temp)
Annual Failure Rate
Warm (above avg. temp)
Barracuda 1.5 TB
(ST31500541AS)
7.9% 11.0%
Barracuda LP 1.5 TB
(ST31500341AS)
15.6% 34.6%

Why is the correlation weak when these numbers look so obvious? It’s because there’s a lot of overlap between the temperatures of the failed drives and the temperatures of the working drives, so you can’t predict for sure which drives will fail, but the low p-value means that there is a meaningful difference in failure rates.

The failure rate of the ST31500541AS does go up at higher temperatures:

blog-temp-seagate

This contrasts with most other drives we have, which don’t show that trend. The Hitachi HDS722020ALA330 is another one of our older drives, and it shows a more typical non-pattern:

blog-temp-hitachi

Seagate Barracuda 3.0TB and Hitachi Deskstar 7K2000

These are the remaining two drives that have a statistically significant correlation between temperature and failures, but they very weak correlations and they are in opposite directions. The Seagates fail very slightly more when they are warmer, while the Hitachi drives fail very slightly more when they are cooler. The correlations of 0.03 and -0.04 are weak enough that we shouldn’t draw conclusions from them.

Hard Drive Temperature Takeaways

Overall, there is not a correlation between operating temperature and failure rates. The one exception is the Seagate Barracuda 1.5TB drives, which fail slightly more when they run warmer.

As long as you run drives well within their allowed range of operating temperatures, keeping them cooler doesn’t matter.

Brian Beach

Brian Beach

Brian has been writing software for three decades at HP Labs, Silicon Graphics, Netscape, TiVo, and now Backblaze. His passion is building things that make life better, like the TiVo DVR and Backblaze Online Backup.
  • Tim Small

    Excellent data thanks… You seem to be operating your datacentre at coolish temperatures by the look of things. Most of the larger operators go up to around 27 Celsius (80F) to reduce energy usage from cooling (higher end of the ASHRAE guidelines) – the Hitachi drive sitting on the desk in front of me is rated up to 70 Celsius IIRC.

    I’m wondering if the drives which run at the warmest temperatures also end up being subject to the most vibration (any correlation between temperature and SMART high-fly writes or hardware ECC corrected variable?) – or indeed if some drives run warmer because of increased mechanical friction in their mechanisms – so perhaps an elevated temperature in an individual drive relative to its near-neighbours (or airflow temperature in the chassis) might be a failure indicator?

  • Andrew Fox

    You say that there is no correlation but in my computer right now my coolest drive is 31 degrees and the warmest 41 degrees. 2 running at 41, 3 running at 37). The room temperature is currently 26 degrees. Do you think 40 degrees is bad temperature for a hard drive?

    • Desolutional

      Generally stay under 45, and always above 25. Too cold and the fluid inside the drive has to work harder to spin the platters (which means more failure), heat isn’t so bad and up to a certain temperature isn’t supposed to affect the drive significantly, rather it actually helps to make the fluid more “fluidy”. My 3TB Seagate drives’ have been running at 40 celcius for over 2 years, with no head parking, have written over 100TB to each, and none have failed as of yet.

      I’d say temperatures up to 55 celcius for short term applications (like a backup) which lasts up to a maximum of 2 hours, is perfectly fine for a HDD, and this is what my External HDD feels best with. I’d never run a drive above 50 celcius for longer than a few hours at a time.

      • Andrew Fox

        I have an external 4TB drive that I told to never spin down in seagate dashboard and it’s running a pretty much constant 45 degrees. Pretty warm, don’t know how I could cool it down though seeing as it is inside an enclosure. 2 of the drives in my PC died last month – one of them was the 40 degrees one that had perfect smart data and the other had a poor smart reading (91% health on Hard Disk Sentinel program) so I had already backed up the data on that one to a new drive thankfully. Do you think SSDs are effected much by temperature?

        • Desolutional

          SSDs are fine up to around 80 degrees C. They won’t heat up during operation however, which makes them good for mini PCs where the case temperature is high.

  • first last

    So far reading a few things here I like the overall format and everything. I do find this article to be a bit misleading though. It seems as though the drives in your pods never go above 38C or around 100F. That has been the defacto temp max (before the life of the drive starts to degrade) that has been going around for a while now, and one I typically use, to save drives from failure. I will agree that so long as you stay within say 75 to 100F window the temperature doesn’t matter. It’s only when you get into the sustained temps of above 100F. I don’t have access to as many drives as you have of course, but I have had access to many drives that were operated at well above 100F before. The longest living drive I have seen that lasted that had an average temp, and or heavily sustained temps of 110F and higher was 3 years, with most failing at around the year and a half mark. Datacenter conditions I assume are a bit better than end user’s computers, with much wider temperature swings, vibrations, etc which could lead to some disparity as well for normal users vs data centers.

    What I would really liked to have seen in this article is max temps reached, and a semblance of sustained high temps, and how that correlated with drive failures, not just the average temp over the life of the drive. I think that might show a higher correlation to drive failures.

    Also you said that the google study said that temperature was not a good indicator. But the study points to higher failure rates when drives are below 25C/77F and higher than 45C/113F. Although what you said is technically true, it’s not necessarily an indicator when a drive will fail, it is a good indicator that temperature can reduce the life span of the drive. Say a drive under optimal temps could last 5 years, lower or higher temps, could reduce that life span down to say 3 or 4. Think of it like for every moment above/below a certain temperature the life span is reduced by x amount, with of course a differential for every degree beyond the max/min.

  • Dan Schwartz

    Two questions:

    1) Are the drives in your “pods” oriented in the horizontal or vertical plane? This makes a major difference as to bearing wear, as when the drives are vertical as both the platter spindle and head arm shafts are horizontal, and do not present an axial load on the thrust bearings, resulting in less friction and hence less heat. In addition, when the drive is oriented vertically, there is also no torque load along the head arm axis, as excessive “play” in this bearing will result first in reduced head flying height (as reported by SMART), and then eventually a head crash into the platter;

    2) If the drives are oriented vertically, what is the orientation of the head arm with respect to the platter spindle, i.e. if the drive is oriented with the long edge down is the force of gravity pulling the arm down or up? This is important as i²R heat generated in the windings of the head arm servo is proportional to the square of the current pulse (i²T [sic] heating). With the drive on it’s long side down and the arm spindle above the platter spindle, gravity balances out the return spring force, requiring about equal current pulse energy levels to move the arm (1²+1²); while if the arm spindle is below the platter spindle, it will require twice the current pulse energy to overcome both the return spring and gravity (2²+0²).

    Dan Schwartz,
    Editor, The Hearing Blog
    http://www.TheHearingBlog.com

    • Emiel Bart

      if you have ever looked a bit more at the website, you will see they mount the drives vertically with the smallest side on the bottom.

  • Tapio Haapala

    How about temperature changes? We have noted (can be statistic error) that after some cooling failures (disk temps peak 29c to 36c) it will cause peak at failures in next 2 weeks. If backblaze guys can confirm or unconfirm this I will be happy :)

  • grenadeh

    These figures don’t mean anything to consumers. All of those averages assume the drives are well cooled. What would be better information is these drives avg temperature in a realistic consumer build. I can tell you for a fact that my Seagate barracudas average 40 celsus with air coolin

  • http://SafeCycling.ca tOM Trottier

    Your temperature range is lower than my external 1tb Hitachi HD. It idles at 40C & ran at 57C until I changed the acoustic and power management to use the minimum power. Now it runs around 50 to 51C. It is specified for operation up to 60 degrees Celsius. I believe the Google report showed that hard drives running under 40 or over 50 were more in danger of failing. But none of your drives run over 40 at all. If you should try running one set of drives at the level of 40 to 50°C.