Almost a year ago, we covered Backblaze’s decision to phase out Seagate 3TB drives after seeing unacceptably high failure rates from one drive in particular, the ST3000DM001. Now, Seagate is facing a class action lawsuits brought on behalf of its customers who bought that particular model — and Blackblaze’s data is mentioned prominently in the suit.
Backblaze’s failure data as of April 2015
Seagate failure rates on the ST3000DM001 weren’t just far higher than other drives, they were also distributed differently. Normally, products follow what’s called a “bathtub curve” failure rate. That means an initial high period of failure as defective units die, followed by low overall failure rates until end-of-life, when hardware begins burning out. The Seagate drives in Backblaze’s storage pods did not exhibit this type of curve.
Backblaze 3TB failure data as of 2015
As of April 2015, just 6% of the original 3TB drives Backblaze purchased were still in service. This kind of data could be evidence of a significant problem with the drive family.
The complaint
The complaint notes that Seagate’s ST3000DM001 was the first 1TB drive to use three platters at 1TB each. This is in contrast to other 3TB drives then on the market, which used 4-5 platters to hit their 3TB densities.
The company made a number of marketing claims that emphasized the reliability of the ST3000DM001, including the claim that the annualized failure rate of the drive is less than 1% and non-recoverable read failure rate is extremely small. These figures are both highly suspect when used to calculate overall drive reliability, but they’re the only information that a hard drive manufacturer will typically release.
The class action suit leans heavily on the Backblaze report — and that’s where problems may arise.
Does Backblaze’s data accurately capture the failure rate?
The class action suit does offer the example of a plaintiff who purchased a drive, experienced an early failure, and then replaced it again with a warrantied drive that also failed. It also leans on the Backblaze information on the ST3000DM001 to support allegations that the drive family was broken.
This argument will hinge on whether Backblaze’s use of the drives in a commercial storage pod constitutes a proper environment for representative testing. I suspect Seagate will argue it does not. Seagate manufactures enterprise-class drives that are specifically designed for reliable operations in challenging environments, and the company will likely claim that the reason Backblaze saw such high failure rates on the ST3000DM001 is because they operated the drive incorrectly.
The fact that the Seagate drives failed in huge numbers while competitor drives did not could be evidence of a defect across the entire product line, or it could simply mean that the other consumer drives are over-engineered.
Death Star vs. Failacuda
If Seagate is smart, it’ll examine what IBM did during the 75GXP debacle — then do the opposite. For those of you who don’t remember: Once upon a time, IBM had a thriving HDD business. IBM Deskstar drives had a reputation for stability and reliability, so the launch of IBM’s 75GXP HDD, and the so-called “fairy dust” that IBM sprinkled on the drive to improve its density were both well received.
Failed drive heads on the infamous 75GXP. The drives acquired the nickname “Death Stars” due to their failure rates
Unfortunately, the honeymoon didn’t last. Customers quickly began to complain, first about early failures, then about repeated failures on replacement drives. As problems spread, publications tried to get in touch with IBM to discuss the issues. IBM’s official policy was to tell everyone, including the press, that it would honor the drive’s warranty and nothing else.
Whatever Big Blue thought it was accomplishing, it didn’t work. With no information on the size or scope of the problem, most publications publicly yanked their previous Deskstar endorsements. IBM wouldn’t tell anyone if the problems on the 75GXP also affected the 60GXP families, so customers steered clear of anything with an IBM label.
It’s hard to measure how much financial impact the disaster had on IBM, since it happened synonymously with the first dot-com crash, but the endless stream of uncertainty and overall negative PR did the company no favors. In the end, IBM sold its HDD line to Hitachi.
There’s little risk of Seagate taking a step that drastic — storage was a side business for IBM, but it’s Seagate’s entire focus. With that said, this is a situation that will be better served by frank admissions and service, not by stonewalling and subterfuge.
Post your thoughts below in the comment section.
SOURCE: Extreme Tech
