I am a consultant, and my clients use and buy
a lot of computers and drives. Your experience is
common. There are 4 major makers and 5th place in
volume Samsung. Don't buy from any but those 5,
because the other makers are either 100% refurb
contractors, or low quality new production.
All 5 of them have had poor reliability on new production
since about 2001 or so. I think they were so busy
tripping over each other to accomodate video and mp3
collector driven size demand, that they forgot about
reliability in the storage increases from 2001 to 2004
or so. I did say I would avoid the 7200.7 series you had
problems in. But I don't think it is any worse than the
Maxtor, Hitachi-IBM or WD products that are comparable in
time of manufacture. I THINK the 7200.8 will turn out
to be better, but I can't be sure because it hasn't
been in the field long enough yet. Manufacture's
quoted MTBF are just plain lies. Also, manufacture's
accelerated aging tests (200 drives running hard in
a hot and humid room) are not a great predictor of
real failure rate in the field.
It is an impossible prediction, because by the time
you know the 3 years-in-use percentage-that-died,
the drives are basically technologically obsolete so
all you can do with that number is see which of your
drives you should get your data off of and retire early.
Most people have strong feelings about a certain
brand's reliability or unreliability that is based
on one or two personal experiences, often very painful
experiences. But these experiences lack statistical
merit, and are misleading. A high failure rate generally
affects only a certain model family, and does not give
rationality to any brand-wide loyalty. In my business, I have
found some sources of actual field failure return rates, and
study those. Also Seagate recently changed from a 3 year
warranty to a 5 year warranty, while Maxtor, Hitachi and WD
have changed from a 3 year warranty to a 1 year
warranty (exception: Raptor, a loud small fast drive).
This may imply Seagate has confidence in new production,
or merely that their marketing department has picked up
on user disatisfaction with high failure rates. I would
buy drives with a 30 day warranty over a 5 year
warranty if I thought statistics indicated the 30 day
drive model would be less likely to fail during the
4 years I will use it. I wrote my own software to
do automatic backups data disk to disk during the day.
But still, when I loose a drive, it takes up a lot of
human time. And many people loose data. By comparison,
the hard drive itself is cheap, and the swap for
a blank refurb drive is small compensation. And I don't
like to use the refurb they send back, because
refurb drives have a much higher future failure rate.
About drives with dead circuit boards...
One of my clients had some important data on a drive
that he had foolishly stopped backing up. Maybe not
important enough for a $5000 fee data recovery company,
but pretty important. Since it was silent and no longer
recognized by the BIOS, I bought a same-model same-year
drive off ebay, and swapped circuit boards. I was
able to recover all his data that way, very cheaply,
making my client very happy. On this model, it worked.
But it is possible some models might use EEPROM
or some other on-circuit-board storage for something
like mapping out defects. But tradtionally, the defect
map is on a special reserved place on the drive media,
so the swap of circuit board worked fine. Typically a
circuit board failure causes your drive to stop being
recognized at all by the BIOS, and no unusual noises
eminate from the dead drive. Typically a mechanical
failure causes "new" noises, errors logged in the
XP / Win2000 / Win2003 "Event Log", or rapid increases
in the "Relocated Sector Count", and "Read Error Rate",
which for PATA-IDE drives can be observed anytime
with the lovely and free program Diskcheckup.exe.
If you know how to read this in SATA, please advice.
Another sign is unusally long delays when copying
files (many retries, and Windows says nothing about
this). Watching these 4 things (noises, Diskcheckup.exe,
and Control Panel | Admin Tools | Event Log, riduculous
delays in file copies) I have been able to replace
about 90% of the ailing drives on my machines or
supported client's machines, while all their data
could still be read (although sometimes slower due to
retries).