The operational use of a drive also plays a great deal into its expected life. Server banks versus PC usage are dramatically different. Google published a similar, more comprehensive,
study in 2007, which was also based on server usage of drives, but they also defined "failure" as "the most accurate definition we can present of a failure event for our study is:
a drive is considered to have failed if it was replaced as part of a repairs procedure." People running servers treat drives as expendable, so at the first sign of "anything," they will swap it out (i.e. "fail" it) just to be safe and keep server banks online.
There are a lot of nuances that can play into "reliability" of a HDD, so the operation and maintenance of the drive also take a large part in this. Many of the drives mentioned were not designed specifically for server usage, they were designed for desktop usage. Very different environments, so take papers like this with a grain of salt.
I don't want to delve too deeply into this, but the big thing to steer clear of has already been mentioned in many threads... "green," low RPM, and drives not SATA III stay clear of... the reason the heads in those drives are there is because they do function, just not at the data rates needed for 7200+rpm.
Edit: As far as SSDs, the above recommendation is sound (Samsung 840 Pro Series)... check out the
benchmarks of the top SSDs
and the price point at the right!!
Shop wisely!! (or risk paying 20 times for something half as fast)