IBM never did fix the problems - they just sold their hard drive division to Hitachi instead.
Over the past couple of years I've seen more dead IBM deskstars than every other make of hard drive combined. They are awful devices with the life-span of a fruit fly. :-(
IBM's division made a lot of drives. A
lot of drives. Some models were/are wonderful, some utter crap. Remember, in a company as huge as IBM, they had separate teams working on RAID-purposed mechanisms, low-cost desktops, portable/micro-drives, and so on.
Rumor suggests the 'Deathstars' - those awful, fruit-fly-lived GXPs - were, in part, a victim of design by committee. (At least, if there's any truth at all to the rumor that the disk controller couldn't keep up with the bitrates off the denser platters as they launched the 60 and 120GB models - leading to dropped bits, including the various internal control signals encoded on the platters, which would then lead to wonderful thunking, head crashing, etc.)
Of course, it's very hard to know how to pick a winner - IBM certainly certified even those bad drives for use in 'reliable' configurations - but it's good to know what you're up against. IBM, and presumably now Hitachi, are prone to these sort of lemon-model problems - same goes for Seagate, perhaps, with their wide array of drives... Contrast to Fujitsu's recent debacle, traced reputably (as in, not a rumor, it's generally accepted) to the use of a bad potting epoxy/ceramic on certain critical chips (possibly the fault of a third-party?)... or Western Digital, who, from anecdotal evidence, just seem prone to quality control problems across the board.
It's a shame the Deathstars sucked, it certainly hurt IBM's reputation (and between that and Fujitsu, it's made it that much harder to pick a manufacturer you could call 'reliable'), and they certainly handled it poorly. Based on organization and 'philosophy,' I'd say Seagate is equally at risk for that sort of problem (though, since they haven't had one recently, I'd say their prerelease testing and QA have thus far been working better than IBM's did), while Quantum/Maxtor, with their smaller lineup- sharing more parts across more models- seems potentially more prone to WD-style 'across-the-board' issues. (That said, I've not had any firsthand experience with Maxtor since the 500MB era.)
I try to like Fujitsu, and I don't think anyone could've predicted their trouble (whether the chips were produced internally or by a third party... who'd expect bad packaging on ICs in the 21st century? We were supposed to have figured that out ten years ago!)... Shame they shot the food when it came time to recall/replace, too.
But like I'm saying, the best you can do is try to have a 'feel' for the companies, make sure nobody else is already outright complaining about the particular model you're considering, and try to gauge your risks appropriately. No drive lives forever, unless it's a Micropolis, in which case it was probably stillborn.
