Hrm.. has it ever been proven that "enterprise level" hard drives crash less than others?
It's not as though they put in a magic circuit that fixes the things that total most drives, and if they did, it wouldn't cost more to put them in the "consumer" drives.
In my experience, Yes, the enterprise grade disks do last longer (or can handle a heavier beating err... load ... than a consumer unit for the same period of time, such as what typically occurs with RAID use).
In the past, an enterprise disk would typically use better mechanical components (i.e. better servos, spindle motors, bearings with tighter tolerances, and better platter assy mounting methods to keep vibration in check). Unfortunately, this isn't really the case any longer (they do still "cherry pick" platters to meet 1E15 Bit Error rates for the enterprise units, and keep consumer disks at 1E14), but it's now the same mechanical parts with additional sensors (i.e. fly height adjustment that is used to combat platter vibration). The firmware also changes to utilize the additional sensors (and why enterprise versions are stable when attached to RAID cards, as the recovery timings are adjusted for this type of use).
Some makers, such as Hitachi, have had their consumer units suffer in quality due to a manufacturing facility relocation (to China in Hitachi's case), which is reflected in the RMA filings (newegg can give a clue to this, though as you pointed out, how the reviewer considers other factors can generate a false negative or positive, such as 4 eggs for a DOA disk that was replaced quickly by newegg). Simply put, results such as this is a result of cost cutting gone awry.
I look at the stats from Newegg's review section and they're all over the place. Consumer and "enterprise" drives have ratings that are usually the same. In fact, I read some of the 4 star reviews of WD Blacks, for instance, and they're a LOT less likely to rate based on failure, skewing the stats even more. There are a few reviews that say "well the first one I got failed, but since I got a new one from WD quickly, I'm giving this 4 stars."
No, the results aren't absolute due to such reviews (i.e. they gave a good rating on a DOA disk due to newegg's handling of a return), but it can give a good clue as to the defect rate (it includes damage that can occur during the shipping, but it's ultimately relevant IMO, as it's part of the real world).
But to get a clearer picture (eliminate such ratings), it means reading all the reviews (at least a decent scan), and recompute them (i.e. ding a DOA or early failed disk to 1 egg, no matter how well newegg handled the problem and was subsequently rated by the original poster). Then calculate the 1 egg reviews in terms of %. It's the closest thing you can do with recent disks, as long term data (i.e. what gets generated by disk recovery services) isn't available until years later.
In addition, Google did some hard drive tests a while back and released the data. They found "consumer" drives are incredibly reliable.
If this is what I think you're referring to, that's long term data generated by the disk recovery industry, so the data is based on older disks.
Aggressive cost cutting may have taken place since those units were made, and has had a negative effect on the products as a result.
I switched to a 10k RPM raptor years ago and noticed a HUGE difference with OS X 10.3 and 10.4. However, I retired the drive for a while and dug it up a couple months ago to put 10.6 on it.. I don't notice much difference at all now between the drive I was using (a 750GB WD) and the 10k.
Platter density is a major focus of the mechanical storage industry, and is where most of the speed increases comes from. So a few years makes quite a difference, and why an earlier Raptor or Velociraptor can no longer beat the pants off of more recently produced 7200 rpm disks (if the older 10k rpm units even beats it at all).
I suppose I could clock the boot times and see which is better, but I have a bunch of drivers that would probably handicap my established installation unless I cloned it.
Test the disks for random access performance, as that's what booting relys on.
No need to clone, set the disk to be tested to the boot location, ... to do a stopwatch test.
