Google Releases Paper on Disk Reliability

[KenR]

Troubling news about the effectiveness of SMART

Despite this high correlation, we conclude that models based on SMART parameters alone are unlikely to be useful for predicting individual drive failures. Surprisingly, we found that temperature and activity levels were much less correlated with drive failures than previously reported...

http://hardware.slashdot.org/hardware/07/02/18/0420247.shtml

[mouser]

The conclusions are interesting that heat and usage is not a very good predictor of failure.. I guess basically the take-home message is what we all suspect: failure of disk drives is unpredictable - so backup your stuff regularly!

[Carol Haynes]

Also mentioned on the BBC website: http://news.bbc.co.uk/1/hi/technology/6376021.stm

[f0dder]

The conclusions are interesting that heat and usage is not a very good predictor of failure.. I guess basically the take-home message is what we all suspect: failure of disk drives is unpredictable - so backup your stuff regularly!

Heat and usage might not be absolutes, but in my humble experience with a much more limited amount of drives than google , excessive heat and the kind of disk usage where there's a lot of head movement (ie, fragmented files as opposed to long linear unfragmented) reads/writes) do wear down drives faster.

[app103]

Here is an article comparing the Google results to the recent Carnegie Mellon study:

http://www.pcworld.com/article/id,131168-page,1/article.html